CN108733952B - A three-dimensional characterization method for spatial variability of soil water content based on sequential simulation - Google Patents

Abstract

The invention discloses a three-dimensional representation method of soil water content spatial variability based on sequential simulation, which selects exemplary soil plots, divides the soil plots into a plurality of grids, and sets random sampling points in each grid; Measure soil water content data at all random sampling points, and perform normality test on soil water content data at random sampling points; perform multi-layer process processing on each grid to obtain multi-layer fine grids; adopt sequential simulation method Simulate multi-layer fine grids, including random walk, local search and condition estimation, to obtain 3D volume data of soil water content; use GRID and TIN data formats to alternately express 3D volume data of soil water content to form soil water content 3D model. The invention obtains the continuous surface of soil water content based on the sequential simulation method, and uses the GRID or TIN format to express the soil water content in three dimensions on the basis of the continuous surface, which overcomes the smoothing effect of the Kriging method.

Description

A three-dimensional characterization method for spatial variability of soil water content based on sequential simulation

technical field

The invention relates to a three-dimensional characterization method of soil water content spatial variability based on sequential simulation.

Background technique

Soil water is the basis for the survival of plants in the ecosystem, and it is also the most active part of the water cycle in the basin, affecting plant growth, the ecological environment and the rational distribution and efficient use of water resources. The soil moisture variable profoundly affects the physical properties of the soil, the transmission and migration of soil nutrients, and plays a very important role in the growth of crops and water-saving irrigation. Therefore, special attention should be paid to soil moisture conservation in the process of crop growth to improve the soil environment for crop growth. According to the soil moisture content and crop water demand during different growth periods of crops, precise irrigation should be implemented to save water resources and improve water resources utilization efficiency. In the actual research process, it is usually impossible to measure the soil water content of each sampling point in the study area one by one. Generally, some discrete sample points are selected for measurement, and the value of soil water content at the unsampled points is obtained through a mathematical model, and a seamless point-to-surface expansion surface is obtained. The sampling points can be randomly selected, layered, or regularly selected.

For the method of extending discrete points to surfaces, extensive research has been carried out at home and abroad. There are many methods to generate raster surfaces by spatial interpolation of point data, such as global polynomial interpolation, inverse distance weight, radial basis function, improved Sheppard method, Kriging, natural neighborhood method, spline function method, etc. Each method has certain assumptions when forecasting and estimating according to the regional variables to be modeled and the conditions of the sampling points. The effect of area interpolation is better. In recent years, the combination of geostatistics and classical statistics with Kriging as the core has gradually become a feasible and widely used method for describing and analyzing regionalized variables. This method is widely used in various fields and involves a wide range of content. It is not only used for soil water content, soil basic physical and chemical properties, soil temperature, biological properties, heavy metal pollution, but also for crop productivity simulation and prediction, crop evapotranspiration, crop growth and development stages, and crop growth traits. It is also widely used in economic and social fields such as the spatialization of agricultural output. The point-to-area extension study of regionalized variables covers different scales, such as farm scale, farmland scale, small scale, county scale, and global scale. The premise of the Kriging method is that the distance and direction between the sampling points can reflect a certain spatial correlation, and use them to explain the spatial variation. Kriging uses a certain mathematical function to fit a specific point or all points within a given search radius to estimate each point. point value. The application of this method realizes the expansion and expression of point data to surface data, but the method itself is prone to a smoothing effect, which smoothes the maximum value in the estimation process, which reduces the estimation accuracy of unsampled points.

In traditional applications, the focus of establishing an uncertainty model for valuation is to determine the mean and variance, which is obviously an incomplete description of the uncertainty of random variables. A more complete description of the statistical uncertainty at a point entails estimating the probability distribution of the modeled variables. In fact, the estimated probability distribution relies on the collection of samples near the estimated point or other known information (such as soil type in the agricultural farming process, the level of cultivated land, management measures, and farmers' comprehensive empirical measures to improve yield, etc.) , knowledge and historical data). How to quantify the point-to-surface representation of the aggregated information of a sample is itself a very complex scientific problem. Geostatistics with Kriging as the core combines spatial analysis technology for spatial interpolation calculation. Although the spatial expansion of regional variables is realized, the estimation of this spatial surface has a smoothing effect. Around the accuracy of spatial estimation of regionalized variables, researchers have made various attempts. Some researchers use multi-point geostatistics to estimate and predict regionalized variables in space, and multi-point geostatistics simulates the training of multiple points. The image replaces the variogram, which more effectively reflects the spatial distribution structure of the research target. Other researchers put forward the random field theory and the transition probability function (Transiogram) theory of the horse chain, and based on this, they constructed the theoretical framework of the horse chain geostatistics, and proposed the linear interpolation method and Mathematical model simulation method.

A typical example of sampling point Kriging interpolation is to use a set of sampling points to generate a continuous surface of regionalized variables. The soil moisture value of each sampling point can be obtained in various ways. The soil moisture content can be measured in the laboratory by collecting undisturbed soil in the field. , the soil water content information can also be collected intelligently through the agricultural Internet of Things system, and can also be obtained by measuring instruments such as Time Domain Reflectometry (TDR for short, the same below), and the water content values of other points in the area can be measured. It is derived by Kriging interpolation. Geostatistics is an important mathematical analysis tool for the continuous distribution pattern of spatial variation of regional variables such as soil water content. With the Kriging method and its variants, the co-Kriging and indicating Kriging as the core, the spatial continuous surface expansion of geostatistical regional variables has been obtained. Great progress. However, due to the inevitable smoothing effect of geostatistics with the Kriging method as the core, the estimation and prediction of regional variables such as soil water content deviate from reality to varying degrees. The smoothing effect cannot be solved by the Kriging method itself. In-depth analysis shows that the geostatistics with Kriging interpolation as the core still inevitably has a smoothing effect, and the estimated value cannot reflect the real variation characteristics of regional variables in space. Many researchers have tried to correct this smoothing effect, and some researchers hope to introduce random parameters or multi-model fusion to study the spatial and temporal variation of regional variables, such as introducing random parameters into deterministic models to carry out surface soil saturation at the field scale. Research on the effect of spatial variation of hydraulic conductivity on water leakage in farmland. Combining principal component analysis and ordinary Kriging to carry out research on the spatial variation of soil water content, the results show that the combination of random parameters and multi-models has improved the quantitative expression of spatial variation patterns of regional variables, but the introduction of random parameters requires Certain preconditions are meaningful, which undoubtedly increases the challenge of research.

To sum up, in the prior art, there is still no effective solution to the problem that the smoothing effect produced by Kriging and its derivative methods on point-to-surface estimation cannot reproduce the corrected spatial correlation.

SUMMARY OF THE INVENTION

In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a three-dimensional representation method of soil water content spatial variability based on sequential simulation. Using GRID or TIN format for three-dimensional expression of soil water content, overcoming the smoothing effect.

The technical scheme adopted in the present invention is:

The first object of the present invention is to provide a three-dimensional characterization method of soil water content spatial variability based on sequential simulation, the method comprising the following steps:

Step 1: Select an exemplary soil plot, divide the soil plot into multiple grids, and set random sampling points in each grid;

Step 2: Measure the soil water content data of all random sampling points, and perform a normality test on the soil water content data of the random sampling points;

Step 3: Multi-layer process processing is performed on the grid of each random sampling point to obtain a multi-layer fine grid;

Step 4: The multi-layer fine grid is processed by the sequential simulation method, including random walk, local search and condition estimation, and the three-dimensional surface data of soil water content is obtained;

Step 5: Use the GRID and TIN data formats to alternately express the three-dimensional surface data of soil water content to form a three-dimensional model of soil water content.

Further, in the step 2, the method for measuring the soil water content data of all random sampling points is:

Take soil samples on the soil surface at all random sampling points, and measure soil moisture data for the soil samples; or,

Measure soil water content data at all random sampling points by a curing instrument; or,

The soil water content data of all random sampling points were measured by the agricultural IoT system.

Further, in the step 2, the step of performing a normality test on the soil water content data of the random sampling points includes:

Perform normality test on the soil water content data of the random sampling point. If the soil water content data of the random sampling point is normally distributed data, then go to step 3. If the soil water content data of the random sampling point is not normally distributed data, perform data transformations, including taking logarithms, sine or cosine.

Further, the step of performing multi-layer process processing on the grid of each random sampling point to obtain a multi-layer fine grid includes:

The grid of each random sampling point is sampled by the diagonal sampling method, the plum blossom sampling method, the checkerboard sampling method or the S-shaped sampling method, and several sampling points are obtained;

According to the set grid cell size, the sampling points are rasterized to obtain a multi-layer fine grid.

Further, the step 3 further includes screening a variogram model, and the variogram model includes an exponential model, a Gaussian model or a spherical model.

Further, the sequential simulation method is used to process the multi-layer fine grid, including random walk, local search and condition estimation, and the step of obtaining the continuous surface of soil water content includes:

Select a random position X in the fine grid, and determine all the nearest neighbor positions within the set search radius of the X position;

Based on the variogram model, obtain a predicted value of X and the estimated standard deviation at X as a linear weighted combination of N selection points;

Using the cumulative normal distribution of the mean M and the standard deviation SD of the predicted value of the random position X of the soil water content at the N points, screen a random variable and use it as the estimate of X;

Next, select another random position X in the fine grid, obtain the random variables of the random points of the fine grid according to the above method, and cycle in turn until the random variables of all random points of the multi-layer fine grid are obtained;

According to the random variable data of all random points, the three-dimensional surface data of soil water content is obtained.

Further, the GRID data format refers to a data format in which the three-dimensional distribution of soil water content is represented by a regular array, and each data in the data format represents an attribute characteristic of soil water content.

Further, the TIN data format is a high and low change value of soil water content.

The second object of the present invention is to provide a computer device for three-dimensional characterization of soil water content spatial variability, comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the following steps are implemented, including:

Select an exemplary soil plot, divide the soil plot into multiple grids, and set random sampling points for each grid;

Measure the soil water content data of all random sampling points, and perform normality test on the soil water content data of random sampling points;

Multi-layer process processing is performed on the grid of each random sampling point to obtain a multi-layer fine grid;

The multi-layer fine grid is processed by sequential simulation method, including random walk, local search and condition estimation, and the three-dimensional surface data of soil water content is obtained;

Using the GRID and TIN data formats to alternately express the three-dimensional surface data of soil water content to form a three-dimensional model of soil water content.

The third object of the present invention is to provide a computer-readable storage medium on which a computer program for three-dimensional representation of spatial variability of soil water content is stored, characterized in that, when the program is executed by a processor, the following steps are implemented:

Select an exemplary soil plot, divide the soil plot into multiple grids, and set random sampling points for each grid;

Measure the soil water content data of all random sampling points, and perform normality test on the soil water content data of random sampling points;

Multi-layer process processing is performed on the grid of each random sampling point to obtain a multi-layer fine grid;

The multi-layer fine grid is processed by sequential simulation method, including random walk, local search and condition estimation, and the three-dimensional surface data of soil water content is obtained;

Using the GRID and TIN data formats to alternately express the three-dimensional surface data of soil water content to form a three-dimensional model of soil water content.

Compared with the prior art, the beneficial effects of the present invention are:

(1) The present invention adopts the sequential simulation method to obtain the continuous surface of soil water content, and uses the GRID or TIN format to express the soil water content in 3D on the basis of the continuous surface, which can correctly and effectively express the continuous surface of soil water content. , and stored in different data formats, it can continuously generate 3D surfaces under different data, and form a 3D series of graphs showing the dynamic changes of soil water content. , On-demand irrigation provides a basic basis, and variable irrigation can be carried out in a targeted manner, which can save water and improve economic benefits;

(2) The present invention adopts the sequential simulation method to effectively deal with the anisotropy of soil water content, overcomes the smoothing effect of the Kriging method, and expresses the specific spatial pattern quantified by the variogram or histogram through a series of random simulations, improving the estimation Mild, avoids estimation distortion caused by "smoothing" effects.

Description of drawings

The accompanying drawings that form a part of the present application are used to provide further understanding of the present application, and the schematic embodiments and descriptions of the present application are used to explain the present application and do not constitute improper limitations on the present application.

Figure 1 is a flow chart of the three-dimensional characterization method of soil water content spatial variability based on sequential simulation;

Fig. 2 is a schematic diagram of a spherical model of soil water content;

Figure 3 is a three-dimensional model diagram of soil water content.

Detailed ways

The present invention will be further described below with reference to the accompanying drawings and embodiments.

It should be noted that the following detailed description is exemplary and intended to provide further explanation of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

It should be noted that the terminology used herein is for the purpose of describing specific embodiments only, and is not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly dictates otherwise, the singular is intended to include the plural as well, furthermore, it is to be understood that when the terms "comprising" and/or "including" are used in this specification, it indicates that There are features, steps, operations, devices, components and/or combinations thereof.

As described in the background art, methods such as Kriging in the prior art have low estimation accuracy, and there is a "smoothing" effect, which is likely to cause estimation distortion. In order to solve the above technical problems, the present application proposes a sequential simulation-based soil A three-dimensional characterization method for spatial variability of water volume.

As shown in FIG. 1 , an embodiment of the present invention provides a three-dimensional characterization method for soil water content spatial variability based on sequential simulation, and the method includes the following steps:

Step 1: Select an exemplary soil plot, divide the soil plot into several grids, and set random sampling points for each grid.

The number of grids is determined according to different research purposes.

Step 2: Measure the soil water content of all random sampling points, and perform normality analysis and test on the soil water content data of random sampling points.

At random sampling points, soil water content is obtained through experiments by obtaining soil samples, or the soil water content of random sampling points is measured by solidifying instruments such as TDR, or the information collection of soil water content at random sampling points is completed through the agricultural Internet of Things system.

Then perform normality test on the soil water content data of the random sampling point. If the soil water content data of the random sampling point is normally distributed data, then go to step 3. If the soil water content data of the random sampling point is not normally distributed data, then perform data conversion, and the models for data conversion include logarithm, sine, cosine, etc. At the beginning of the calculation, complete the point data normality test of soil water content.

Step 3: Perform multi-layer process processing on the grids involved in random sampling points.

In order to enhance the improvement of long-range effects and short-range variability, a multi-layer process is performed on the grids involved in random sampling points, starting from a relatively rough grid, gradually generating finer and finer grids, and finally forming A fine grid.

The multi-layer processing process for the grid of each random sampling point involves two aspects. One is the sampling method. There are many sampling methods. The sampling methods are different in different situations. They are: 1. Diagonal sampling method: Suitable for sewage irrigation plots, sampling at the center point of each diagonal line; 2. Plum blossom sampling method: suitable for plots with small area, flat terrain and uniform soil; 3. Checkerboard sampling method: suitable for medium 4. S-shaped or serpentine sampling method: suitable for smaller areas, less flat terrain, less uniform soil, and more sampling points must be taken 5. Grid sampling method, the black dots in the three-dimensional image of the patent of the present invention are grid sampling; the other aspect refers to the grid processing of sampling points. Because the data structure has a typical matrix structure, the computer Time-saving calculation; Cell Size, also known as analysis resolution, can be set artificially according to research purposes. For example, when performing fusion analysis with satellite images, it is generally consistent with the resolution of satellite images, such as remote sensing. The resolution of the TM image is 25m*25m; the spatial analysis of raster data is carried out on the basis of each raster unit. If the raster unit is too large, the accuracy of the analysis results will be reduced, and if the unit is too small, a large number of data, and the calculation speed is reduced. Therefore, it is necessary to select the appropriate unit size.

The multi-layer processing process of the grid of random sampling points in the present invention is the process of setting appropriate grids and subdividing according to different research objectives. The finer the multi-layering of the grid, the more time-consuming the calculation is. Therefore, the realization process of the grid multi-layer process of random sampling points is not complicated. The key is to set the size of grid cells according to the research purpose, and then select the number of grids to perform computer automation processing.

Screening variation function models, screening variation models are mainly based on curve fitting for automatic computer processing and operations. The variogram model mainly includes a variety of models such as exponential model, Gaussian model or spherical model. Figure 2 shows the spherical model of soil water content.

Step 4: Sequential simulation of soil water content at random sampling points based on random walk, local search and conditional estimation to form a seamless surface.

Sequential simulation integrates various information by stochastic modeling of soil water content attributes, and integrates the correlation and uncertainty of these information into the model, emphasizing that the probabilistic model is the overall probabilistic feature of action and outcome. In the process of sequential simulation, it is necessary to convert the simulated value conditions of the measured points into the actual measured value, so the simulated value of the measured point is equal to the measured value, which is suitable for quantitatively characterizing the heterogeneity and uncertainty of soil water content. With the increase of the number of simulations, the sequential simulations describe the distribution of soil parameter values in more detail in the entire simulation area, the ratio of nugget value to sill value gradually increases, and the variation range gradually approaches the measured data.

Based on random walk, local search and conditional estimation, the specific implementation method to realize the sequential simulation of soil water content at random sampling points to form a seamless surface is as follows:

For the random position X in the fine grid, determine all the nearest neighbor positions located at the X position and within the defined search radius; That is to say, with the change of distance, the soil water content itself presents the relevant characteristics with the change of distance.

Using the variogram model obtained in step 3 to provide one predicted value at X and the estimated standard deviation at X, as a linearly weighted pool of N selection points, and using a cumulative normal distribution with mean M and standard deviation SD, filter A random value of a random variable to be used as an estimate of X. As a linearly weighted combination of N selected points can be considered as an indicator, the mean M is the average of the predicted values at random locations X of soil moisture content over N points.

Then randomly walk to another unvisited grid position, and continue to determine random variables according to the definition of the random position of the fine grid and the variogram model, until all the nodes of the fine grid are visited. Repeat this process for multiple layers of fine grids, each time the result plus the input data point as the source data point value to perform a random walk simulation. In this repeated process, the random walk method reduces the artifacts when searching the grid. risk, and achieve the stability of model operation. For regional variables such as soil water content, in the process of participating in the simulation and modeling of raster spatial data, stability is one of the goals, and a stable model promotes the gradual promotion of model applications. In the random walk search process, there are two types of stationarity: one is mean stationarity, which assumes that the mean is invariant and independent of location; the other is second-order stationarity related to the covariance function and semivariogram related Internal stability. Second-order stationarity assumes that the covariance of any two points with the same distance and direction is the same, and the covariance is only related to the value of the two points and not to their location.

A stochastic simulation can generate numerous implementations, each exhibiting the same pattern, but in a different way. In a univariate distribution model, the uncertainty of a random variable is calculated by its series of outcomes. Similarly, the output of a series of stochastic simulations is characterized by uncertainty. In the stochastic process, there are specific relevant rules, which can be used to predict and estimate the uncertainty of the prediction results. With the help of sequential simulation, the uncertainty pattern is expressed in the form of random probability. For intrinsic stationarity, the assumption of intrinsic stationarity of random sample point data shows that the variance of any two points with the same distance and direction is the same. The type aspect is necessary, or stationarity is the premise of establishing the variation model.

This process is the process of obtaining 3D surface data from point data, including linear weighting, design and search of the radius range of adjacent positions, and the participation of statistical indicators such as average value and standard deviation. The estimation distortion caused by the effect is improved, and the estimation accuracy is improved.

As a basic simulation method, Gaussian sequential simulation overcomes the smoothing effect of the Kriging method. This simulation needs to first define a joint probability model of all grid point attribute values, and the joint distribution is defined as

F(z ₁ ,z ₂ ,z ₃ ,...,z _N )=Pr(Z(u ₁ )≤z ₁ ,...,Z(u _N )≤z _N )

u ₁ ,u ₂ ,u ₃ ,…,u _N Z(u ₁ )u ₁

Among them, z ₁ , z ₂ , z ₃ ,...,z _N are the measured values of points u ₁ , u ₂ , u ₃ ,..., u _N ; Pr is the probability; Z(u ₁ ) is the simulated value of u ₁ , Generating a sample from this distribution takes into account the spatial correlation between all points.

The main idea of the sequential Gaussian simulation conditional method is to sequentially obtain the conditional cumulative distribution function of each grid node along a random path, and obtain the simulated value from the conditional cumulative distribution function. The method revolves around the randomness and structure of the spatial distribution of the regional variable soil water content, and reproduces the volatility and discreteness of the spatial information of this variable. The obtained simulation data are used together as condition data, and enter the simulation of the next point, so as the simulation progresses, the condition data set will continue to expand. The spatial distribution trend of the sequential simulation results is consistent with the original data. Further analysis confirmed that the simulated and measured values were equal at the measured points.

The invention adopts the sequential simulation method to effectively deal with the anisotropy of soil water content, and realistically expresses the specific spatial pattern quantified by the variogram or histogram through a series of random simulations. The disadvantage of the Kriging method is that the attribute value of the unsampled point is estimated separately, and the correlation between the point and the unknown points that have been estimated before is not considered. Obviously, the Kriging method cannot reproduce the corrected spatial correlation. This is also the reason for the smoothness of its results.

Step 5: Visualize the soil 3D surface using GRID or TIN data format.

Using the GRID and TIN data formats to alternately express the three-dimensional surface data of soil water content to form a three-dimensional model of soil water content. Figure 3 shows the three-dimensional model of soil water content, the sampling points of the grid samples at the black dots, the lower part is the soil water content represented by GRID, and the upper part is the soil water content represented by TIN format, which can be more vivid. And clearly characterize the three-dimensional change of soil water content.

TIN (triangulated irregular network) is a triangulation method. This data format is used to express the 3D shape of soil water content mainly because the data model has data of irregularly distributed points in the storage space, and obtains other parameters such as slope and In terms of accessibility in topographical parameters for fine irrigation, the data format also enables rapid determination of 3D surface discontinuities, such as extremely steep soil moisture content, sampling points with extremely high or low moisture content. But there are some problems with this data format: First, there are many different triangles that may be generated from the same set of points, and there are many different triangle algorithms that can generate a large number of different, "fragmented" triangles, each triangle algorithm It consumes a lot of computational time than the decomposition of the regular space point set. Second, it is difficult to superimpose two data layers with irregular grids, let alone the deep-level information generated by the superposition of soil nutrient content and soil water content information. The GIRD data format avoids the generation of fragmented triangles, and has advantages in overlaying and interacting with other GRID format layers, performing in-depth analysis of spatial data, and mining valuable variable information, such as soil moisture content GRID format information, Regional crop yield GRID format information, soil nutrient content GRID format information... For the superposition and algebraic operations of these information, we can analyze how soil water content and soil nutrient content affect crop yield based on three-dimensional maps, and we can also understand soil The coupling relationship between nutrients and soil water content, and then analyze the surface law of nutrient solute transport. Therefore, the GRID format and the TIN format can play the advantages of each format for the storage of soil water content data for different research and development purposes.

Kriging is a local weighted average interpolation method. It estimates the soil water content value within the sample point range according to the soil water content data of the sample point. Although this method has a smoothing effect, it is not suitable for the prediction and estimation of soil water content. The expression still has a strong reference significance. For the same soil water content data, the array or hierarchical nature of this data format consumes less time than TIN calculation. In addition, the grid data model GRID is also applied to the base fertilizer fertilization prescription map and precise irrigation map during the whole growth period of crops. Whether the grid fertilization amount is less than or greater than 0 is used to determine the amount of fertilization, and the map algebra operation in the GRID expression can be used to determine the irrigation timing and The amount of irrigation water can be effectively adjusted.

The continuous surface of soil water content is obtained by the method based on sequential simulation. On the basis of the continuous surface, the 3D expression of soil water content using GRID or TIN format is mainly obtained through the following steps:

First of all, it is clear that grid data (GRID) and triangular irregular network (TIN) data can be converted into each other, and this conversion depends on different research purposes.

GRID grid data presents simple and intuitive spatial structure characteristics, also known as grid structure (raster or gridcell) or pixel structure (pixel), which refers to the division of the earth's surface into uniform and closely adjacent grid arrays. A grid acts as a cell or pixel, defined by row, column numbers, and contains a code that indicates the pixel's attribute type or magnitude, or simply a pointer to its soil moisture attribute record. Therefore, the grid data structure is a data organization that represents the distribution of spatial objects or phenomena in a regular array, and each data in the organization represents the non-geometric attribute characteristics of the objects or phenomena.

The formation of the TIN data format is created from a variety of vector data sources. Vector data such as point, line and polygon features are used as the data source for creating TINs. For the purpose of the present invention, it can be understood that the Z value is the high and low change value of soil water content.

GRID and TIN formats are easier to handle in the process of geographic information processing. Mainly according to the digitized map format, set the appropriate grid size, obtain the required number of grids, add some simple lines of code or put the In this way, the codes of the two processing formats can be obtained by encapsulating them to form plug-ins.

The three-dimensional characterization method of soil water content spatial variability based on sequential simulation proposed in the embodiment of the present invention is the soil water content basis for sampling points, whether it is the intelligent acquisition of soil water content data by the Internet of Things system, or the soil water content of the sample points monitored by TDR. This sequential simulation technology can correctly and effectively express the continuous surface of soil water content, and achieve high-precision point-to-surface estimation and prediction. And stored in different data formats such as TIN or GRID. The present invention generates a 3D surface of soil water content based on the three-dimensional representation method of soil water content spatial variability based on sequential simulation, and forms a 3D series map of the dynamic change of soil water content, which can reflect and reproduce the difference of soil water content within a region, and is accurate Irrigation and on-demand irrigation provide a basic basis, and variable irrigation can be carried out in a targeted manner, which can save water and improve economic benefits.

Although the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, they do not limit the scope of protection of the present invention. Those skilled in the art should understand that on the basis of the technical solutions of the present invention, those skilled in the art do not need to pay creative work. Various modifications or variations that can be made are still within the protection scope of the present invention.

Claims (8)

1. a three-dimensional characterization method of soil water content spatial variability based on sequential simulation, is characterized in that, comprises the following steps: Step 1: Select an exemplary soil plot, divide the soil plot into multiple grids, and set random sampling points in each grid; Step 2: Measure the soil water content data of all random sampling points, and perform a normality test on the soil water content data of the random sampling points; Step 3: Multi-layer process processing is performed on the grid of each random sampling point to obtain a multi-layer fine grid; Step 4: The multi-layer fine grid is processed by the sequential simulation method, including random walk, local search and condition estimation, and the three-dimensional surface data of soil water content is obtained; Step 5: Use the GRID and TIN data formats to alternately express the three-dimensional surface data of soil water content to form a three-dimensional model of soil water content; The steps of performing multi-layer process processing on the grid of each random sampling point to obtain a multi-layer fine grid include: The grid of each random sampling point is sampled by the diagonal sampling method, the plum blossom sampling method, the checkerboard sampling method or the S-shaped sampling method, and several sampling points are obtained; According to the size of the set grid unit, the sampling points are rasterized to obtain a multi-layer fine grid; The sequential simulation method is used to process the multi-layer fine grid, including random walk, local search and condition estimation, and the steps of obtaining the continuous surface of soil water content include: Select a random position X in the fine grid, and determine all the nearest neighbor positions within the set search radius of the X position; Based on the variogram model, obtain a predicted value of X and the estimated standard deviation at X as a linear weighted combination of N selection points; Using the cumulative normal distribution of the mean M and the standard deviation SD of the predicted value of the random position X of the soil water content at the N points, screen a random variable and use it as the estimate of X; Next, select another random position X in the fine grid, obtain the random variables of the random points of the fine grid according to the above method, and cycle in turn until the random variables of all random points of the multi-layer fine grid are obtained; According to the random variable data of all random points, the three-dimensional surface data of soil water content is obtained. 2. The three-dimensional characterization method of soil water content spatial variability based on sequential simulation according to claim 1, wherein in the step 2, the method for measuring the soil water content data of all random sampling points is: Take soil samples on the soil surface at all random sampling points, and measure soil moisture data for the soil samples; or, Measure soil water content data at all random sampling points by a curing instrument; or, The soil water content data of all random sampling points were measured by the agricultural IoT system. 3. The three-dimensional characterization method of soil water content spatial variability based on sequential simulation according to claim 1, wherein in the step 2, the step of performing normality test on the soil water content data of random sampling points include: Perform normality test on the soil water content data of the random sampling point. If the soil water content data of the random sampling point is normally distributed data, then go to step 3. If the soil water content data of the random sampling point is not normally distributed data, perform data transformations, including taking logarithms, sine or cosine. 4. The three-dimensional characterization method of soil water content spatial variability based on sequential simulation according to claim 1, wherein the step 3 further comprises screening a variogram model, and the variogram model comprises an exponential model, a Gaussian model or spherical model. 5. The three-dimensional characterization method of soil water content spatial variability based on sequential simulation according to claim 1, wherein the GRID data format refers to a data format in which the three-dimensional distribution of soil water content is represented by a regular array, Each data in this data format represents an attribute characteristic of soil water content. 6 . The three-dimensional characterization method of soil water content spatial variability based on sequential simulation according to claim 1 , wherein the TIN data format is a high and low variation value of soil water content. 7 . 7. A computer device for three-dimensional representation of soil water content spatial variability, comprising a memory, a processor and a computer program stored on the memory and running on the processor, wherein the processor executes the program implement the following steps, including: Select an exemplary soil plot, divide the soil plot into multiple grids, and set random sampling points for each grid; Measure the soil water content data of all random sampling points, and perform normality test on the soil water content data of random sampling points; Multi-layer process processing is performed on the grid of each random sampling point to obtain a multi-layer fine grid; The multi-layer fine grid is processed by sequential simulation method, including random walk, local search and condition estimation, and the three-dimensional surface data of soil water content is obtained; Use GRID and TIN data formats to alternately express the three-dimensional surface data of soil water content to form a three-dimensional model of soil water content; The steps of performing multi-layer process processing on the grid of each random sampling point to obtain a multi-layer fine grid include: The grid of each random sampling point is sampled by the diagonal sampling method, the plum blossom sampling method, the checkerboard sampling method or the S-shaped sampling method, and several sampling points are obtained; According to the size of the set grid unit, the sampling points are rasterized to obtain a multi-layer fine grid; The sequential simulation method is used to process the multi-layer fine grid, including random walk, local search and condition estimation, and the steps of obtaining the continuous surface of soil water content include: Select a random position X in the fine grid, and determine all the nearest neighbor positions within the set search radius of the X position; Based on the variogram model, obtain a predicted value of X and the estimated standard deviation at X as a linearly weighted combination of N selection points; Using the cumulative normal distribution of the mean M and the standard deviation SD of the predicted value of the random position X of the soil water content at the N points, a random variable is screened and used as the estimate of X; Next, select another random position X in the fine grid, and obtain the random variables of random points of the fine grid according to the above method, and cycle in turn until the random variables of all random points of the multi-layer fine grid are obtained; According to the random variable data of all random points, the three-dimensional surface data of soil water content is obtained. 8. A computer-readable storage medium on which a computer program for three-dimensional representation of soil water content spatial variability is stored, wherein the program is executed by a processor to achieve the following steps: Select an exemplary soil plot, divide the soil plot into multiple grids, and set random sampling points for each grid; Measure the soil water content data of all random sampling points, and perform normality test on the soil water content data of random sampling points; Multi-layer process processing is performed on the grid of each random sampling point to obtain a multi-layer fine grid; The multi-layer fine grid is processed by sequential simulation method, including random walk, local search and condition estimation, and the three-dimensional surface data of soil water content is obtained; Use GRID and TIN data formats to alternately express the three-dimensional surface data of soil water content to form a three-dimensional model of soil water content; The steps of performing multi-layer process processing on the grid of each random sampling point to obtain a multi-layer fine grid include: The grid of each random sampling point is sampled by the diagonal sampling method, the plum blossom sampling method, the checkerboard sampling method or the S-shaped sampling method, and several sampling points are obtained; According to the size of the set grid unit, the sampling points are rasterized to obtain a multi-layer fine grid; The sequential simulation method is used to process the multi-layer fine grid, including random walk, local search and condition estimation, and the steps of obtaining the continuous surface of soil water content include: Select a random position X in the fine grid, and determine all the nearest neighbor positions within the set search radius of the X position; Based on the variogram model, obtain a predicted value of X and the estimated standard deviation at X as a linearly weighted combination of N selection points; Using the cumulative normal distribution of the mean M and the standard deviation SD of the predicted value of the random position X of the soil water content at the N points, a random variable is screened and used as the estimate of X; Next, select another random position X in the fine grid, and obtain the random variables of random points of the fine grid according to the above method, and cycle in turn until the random variables of all random points of the multi-layer fine grid are obtained; According to the random variable data of all random points, the three-dimensional surface data of soil water content is obtained.

Images