A method for modeling spatial distribution of soil volumetric water content

Abstract

This invention relates to the field of soil environmental monitoring technology, and discloses a method for modeling the spatial distribution of soil volumetric water content. The method includes: acquiring hourly soil water content data at different station levels; extracting valid samples through multi-threshold and consistency checks; calculating daily values ​​for each station; constructing a spatial variation structure constraint function using the daily station values ​​and latitude / longitude coordinates; performing gridded reconstruction; removing local sampling bias; generating high-resolution gridded daily values; co-measuring the gridded daily values ​​with a drought early warning index; performing time-series secondary decomposition to identify local water deficit trends; and outputting tiered early warning commands and dynamic weight compensation values. This invention addresses the challenges of missing spatial continuity and weak heterogeneity representation at discrete stations. Through anomaly analysis and closed-loop parameter correction, it improves the sensitivity of regional drought detection and the signal-to-noise ratio of monitoring products, thereby enhancing the reliability of agricultural drought decision-making.

Description

Technical Field

[0001] This invention relates to the field of soil environmental monitoring technology, and more specifically, to a method for modeling the spatial distribution of soil volumetric water content. Background Technology

[0002] Soil moisture is a core monitoring indicator in fields such as agricultural meteorological monitoring, drought early warning, hydrological forecasting, and farmland irrigation management. Currently, most regions rely on automatic soil moisture monitoring network to collect soil volumetric water content in real time. This data, as the raw monitoring element without conversion to soil hydrological constants, avoids parameter errors introduced by manual observation from the source and constitutes an important foundational element for soil moisture monitoring. Developing spatial distribution models based on this raw volumetric water content data is directly significant for improving the accuracy of regional soil moisture status assessment.

[0003] Current soil moisture monitoring and analysis technologies largely rely on a single-point data processing model centered on discrete monitoring stations. Specifically, for hourly data on the volumetric water content of shallow soil layers (10 cm, 20 cm, and 30 cm) collected by automatic monitoring stations, the current processing procedure typically involves performing quality control steps such as lower threshold testing, upper threshold testing, time consistency testing, and Grubbs' test on the hourly data for each soil layer. Then, the mean of the valid samples from the three soil layers is calculated, and the average of the three soil layers is obtained to obtain the daily value of the shallow soil volumetric water content for that station. This station-scale data processing logic results in monitoring results that can only characterize the soil moisture status at a limited number of station locations, lacking the ability to intuitively depict the continuous spatial distribution characteristics of soil moisture across the entire province. Furthermore, due to the lack of standardized gridded interpolation processing, single-point data cannot accurately represent the spatial differences and anomalies in soil moisture within a region, exhibiting significant limitations in spatial resolution and hindering the quantitative expression of spatiotemporal deviations in soil moisture.

[0004] After acquiring daily data from various stations, existing methods do not further integrate gridded interpolation and anomaly percentage analysis, making it difficult to clearly reveal the spatiotemporal variation characteristics of soil moisture. This limitation results in insufficient support for refined and comprehensive scientific analysis in agricultural production scheduling, meteorological disaster prevention and control, and hydrological resource assessment, failing to fully meet the overall demand for high-precision, comprehensive monitoring and decision-making regarding the spatial distribution of soil moisture.

[0005] Therefore, how to construct a modeling method that can comprehensively reflect the spatial heterogeneity of soil moisture and improve the ability to characterize spatiotemporal changes has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] In view of this, the present invention provides a method for modeling the spatial distribution of soil volumetric water content, which is used to solve the problems in existing soil moisture monitoring, such as the lack of spatial continuity due to the single-point data processing mode of discrete stations, the weak ability to characterize spatial heterogeneity due to the lack of standardized gridded interpolation processing, and the inability to intuitively quantify the spatiotemporal deviation trend of soil moisture through anomaly analysis.

[0007] The first aspect of the soil moisture spatial distribution modeling method is executed by a processing system. This processing system can be an automated spatial analysis device integrated into a provincial agricultural meteorological monitoring platform, or a distributed data processing and logic operation unit, such as a high-performance computing server, a geographic information processing workstation, an embedded microprocessor control module, or a cloud data analysis engine. In other words, this invention does not limit the specific implementation form of the processing system.

[0008] Firstly, a method for modeling the spatial distribution of soil volumetric water content is provided. This method includes: a processing system acquiring hourly data of shallow layered soil volumetric water content from automatic soil moisture monitoring stations within a provincial jurisdiction in real time each day; extracting effective sample data from each station based on preset lower threshold tests, upper threshold tests, time consistency tests, and Grubbs' test algorithms; calculating the average of the hourly effective samples from the three soil layers at each station scale; and then obtaining the daily value of shallow soil volumetric water content for each station; the processing system collecting daily station value data and corresponding latitude and longitude coordinate data from all stations across the province; constructing a spatial variation structure constraint function based on the latitude and longitude coordinate data; and inputting the daily station value data into the spatial variation structure constraint function for gridded reconstruction, removing local sampling bias components caused by uneven spatial distribution of stations or local micro-topography, and extracting the characteristic of continuous spatial distribution across the province. The system collects daily values ​​of high-resolution gridded soil volumetric water content in a spatially distributed state. It then performs a co-metric assessment between these daily gridded values ​​and a preset drought evolution warning index. If the co-metric assessment is lower than a first preset drought level threshold, the system determines that the current soil moisture state is within a normal fluctuation range. If the co-metric assessment is higher than or equal to the first preset drought level threshold but lower than a second preset drought level threshold, the system performs a second temporal dimension decomposition of the daily gridded values ​​to identify local soil moisture deficit caused by periods of continuous low precipitation or abnormal irrigation interception. Based on the assessment results, the system generates tiered warning instruction labels. When the soil water separation difference exceeds the warning level, it outputs a visualization rendering instruction to the color map driven by thematic mapping. The system extracts the temporal characteristics of gridded anomaly fluctuations over continuous monitoring periods, traces the source in reverse, and outputs dynamic weight compensation values ​​for the spatial interpolation model.

[0009] Optionally, the processing system extracts valid sample data from each station based on a preset verification algorithm, including: capturing the original dielectric response pulse signals of the shallow 10 cm, 20 cm, and 30 cm soil layers in real time by time-domain acquisition units deployed at each automatic water station, and converting the instantaneous value of soil volumetric water content of each soil layer by pulse intensity; collecting the hourly numerical boundaries of each soil layer by threshold discrimination algorithm, and using the comparison relationship between the hourly value and the lower threshold of 3.5% and the upper threshold of 52.0% to drive the valid sample screening unit to update the status marker value of the current time; and triggering the multi-soil layer fusion calculation program to perform the daily shallow layer value integration calculation when the boundary time signal is detected.

[0010] Optionally, the processing system collects daily value data of stations and corresponding latitude and longitude coordinate data, including: sending station geometric topology sensing signals to the gridded processing engine through the spatial registration unit, and receiving the projected position vector value of the station under a unified geographic reference frame; collecting daily shallow layer value data and station elevation attribute data of each station through the data assimilation unit, and reconstructing the effective soil moisture attribute feature vector of the station in the current spatial field.

[0011] By employing an architecture of hourly quality control and spatial variability structure modeling, the processing system maintains a balance between interpolation accuracy and real-time data quality even with uneven spatial distribution of monitoring stations, reducing gridded reconstruction bias caused by the accumulation of errors from individual stations. The spatial variability structure constraint function removes local sampling bias components, improving the accuracy of gridded daily values ​​in representing the overall regional soil moisture distribution. Compared to existing monitoring schemes that rely on discrete station numerical averaging or point mapping, this method, through multi-scale verification fusion and second-order temporal decomposition identification, enhances the sensitivity of capturing local moisture anomalies and regional drought conditions. It supports precise spatiotemporal continuous monitoring of soil moisture in high-resolution, full-domain scenarios, reducing the risk of misjudgments in agricultural production decisions.

[0012] In some implementations of the first aspect, the method further includes: the processing system collects spatial distribution density deviation samples of stations under different seasonal precipitation conditions based on the monitoring station network to obtain the background error covariance distribution vector of the current interpolation model; analyzes the spatial characteristic scale of the background error covariance distribution vector according to the statistical optimization algorithm to determine the search radius and azimuth parameters of the interpolation model; decomposes the neighborhood of the grid points to be interpolated into a standard grid background field and a local topographic disturbance field; while keeping the standard grid background field unchanged, the spatial drift component defined by the representative difference of the station site selection target is filtered out from the local topographic disturbance field by the Kriging weighted algorithm; the processed background field and disturbance field are reconstructed point by point to synthesize a set of spatially enhanced grid points for soil moisture. The processing system extracts the spatial interference of the station distribution configuration in real time and dynamically adjusts the interpolation weights, thus removing the signal modulation interference of local environmental heterogeneity on the spatial estimation results.

[0013] In some implementations of the first aspect, the method further includes: the processing system performing adaptive temporal mode decomposition on the spatially enhanced grid set of soil moisture to obtain the intrinsic temporal function components corresponding to the transient excitation of the region by previous concentrated precipitation; extracting the periodic oscillation envelope of the intrinsic temporal function components to calculate the rate of change of soil moisture decay during the continuous time series; vertically slicing the spatially enhanced grid set of soil moisture to extract the local water content differences within the spatial grid at different soil depths; and performing spatiotemporal multidimensional mapping between the decay rate and the local water content differences to construct a coupled feature vector of soil moisture spatial heterogeneity and temporal evolution. By extracting the moisture decay features, the processing system identifies local water content fluctuations caused by the water-holding characteristics of clay, thereby improving the depth of characterization of drought evolution boundaries.

[0014] In some implementations of the first aspect, the method further includes: the processing system acquiring monitoring feedback sample sets covering different landform types and soil textures; establishing a standard soil moisture spatial distribution center vector based on a clustering algorithm to form a soil moisture anomaly pattern feature library; constructing a judgment model based on geographic weighted regression or hybrid heuristic search, and pre-training it using the standard center vector; inputting the real-time acquired soil moisture coupled feature vector into a preset kernel function, mapping it to a high-dimensional geographic attribute space, and constructing an optimal soil moisture state decision hyperplane; and calculating the geometric distance of the feature vector relative to the decision boundaries of each drought level. Through high-dimensional spatial mapping, the processing system expands the linear separability between the normal moisture spatial curve and drought-deficient fluctuations, improving the robustness of drought judgment decisions.

[0015] In some implementations of the first aspect, the method further includes: the processing system extracts the instantaneous spatial gradient flow of the spatially enhanced grid set of soil moisture based on a sliding window mutation operator, and identifies local anomalous abrupt change patches in the spatial domain; performs narrowband directional gradient spectrum analysis on the grid set to extract the differences in moisture distribution on the windward and leeward slopes on both sides of the topographic elevation gradient; and fuses the contrast factor of the local anomalous abrupt change patches with the aspect distribution difference features to construct a second-order refined spatial diagnostic vector. When the contrast factor is within a first preset interval and the aspect difference exhibits an asymmetric distribution, the processing system determines that there is a real local moisture variation signal caused by topographic convergence lines or leakage from irrigation facilities.

[0016] In some implementations of the first aspect, the method further includes: the processing system performing trend evolution analysis on grid anomaly fluctuation deviation samples within a preset continuous monitoring period, extracting the phase displacement trajectory of the feature vector over time; establishing a correlation model between the phase displacement trajectory and the semivariance function parameters within the interpolation algorithm; and converting the displacement amount into parameter correction instruction code data packets based on an adaptive feedback control algorithm, and transmitting them to the central processing unit of the processing system in real time. Through reverse tracing, the processing system achieves real-time correction of the interpolation model parameters, avoiding systematic shifts in the overall monitoring results caused by sensor aging at automatic stations or changes in the climate background.

[0017] Secondly, a soil moisture spatial distribution modeling device is provided. This device can be the aforementioned processing system hardware or an embedded computing module used to perform the functions of the system.

[0018] One possible implementation is that the apparatus includes units for performing the method steps described in the first aspect. For example, the apparatus includes a data acquisition unit (for monitoring data acquisition and early warning command output) and a spatial analysis and processing unit (for grid interpolation modeling and spatiotemporal feature identification).

[0019] Thirdly, a soil moisture spatial distribution modeling device is provided, including a processor that executes computer instructions in a memory, causing the device to perform the soil volumetric water content spatial distribution modeling method of the first aspect.

[0020] In one possible implementation, the device also includes a communication interface for digital signal interaction with an automatic water station data collector, a geographic information database server, and an integrated early warning display terminal.

[0021] Fourthly, a soil moisture spatial distribution modeling device is provided, including a logic circuit and an input / output interface. The input / output interface is used to receive monitoring signals from multiple source stations and output early warning judgments or parameter correction data packets. The logic circuit is used to execute the spatial variation structure modeling and soil moisture spatiotemporal feature identification method of the first aspect.

[0022] Fifthly, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, performs the spatial distribution modeling method of the first aspect.

[0023] In a sixth aspect, a computer program product is provided, comprising instructions that, when executed on a computer, cause the method of the first aspect to be performed.

[0024] In a seventh aspect, a chip system is provided, including a processor, for calling program instructions in memory to implement the multi-scale data fusion, grid field reconstruction and drought evolution feature recognition functions of the first aspect.

[0025] Eighthly, a chip installed in a gridding processing device is provided. This chip reads multi-source site sensor data through a communication interface and drives a processor to execute the online gridding modeling process of the first aspect, outputting spatial distribution map rendering instructions.

[0026] For a description of the beneficial effects of any of the second to eighth aspects, please refer to the description of the first aspect.

[0027] Compared with the prior art, the present invention has the following beneficial effects: This invention provides a method for modeling the spatial distribution of soil volumetric water content. Through a multi-soil-layer quality control process, it achieves real-time dynamic validity screening of raw monitoring data from various stations. By constructing a spatial variation structure constraint function, this invention solves the problem of continuity discontinuity in regional spatial field reconstruction caused by the discrete distribution of stations, transforming the complex heterogeneous distribution process of regional moisture into a computable standardized grid field, thus improving the geospatial resolution of moisture monitoring. Secondly, by introducing anomaly percentage calculation and a temporal secondary decomposition identification mechanism, it can perform fine-level feature extraction for local anomalies caused by meteorological drought accumulation or soil texture differences, effectively identifying spatial representation biases caused by topographic artifacts. Through the synthesis of spatially enhanced grid sets of soil moisture and high-dimensional geographic attribute spatial mapping decisions, this invention maintains a high signal-to-noise ratio in monitoring products under complex local environmental disturbances and background climate fluctuations, enhancing the system's response sensitivity to regional initial drought signals. Finally, through reverse tracing and dynamic weight compensation mechanisms, it achieves closed-loop optimization of interpolation model parameters, solving the lag problem of traditional single-point analysis logic in dealing with changes in automatic station networks or performance variations of monitoring instruments. This invention not only ensures the spatial continuity of provincial regional water monitoring, but also optimizes the gradient expression within the drought evolution path through spatiotemporal coupling feature analysis, eliminating systematic errors such as uneven station density and sample bias, and significantly improving the quality, reliability, and decision-making efficiency of agricultural meteorological disaster monitoring. Attached Figure Description

[0028] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a method for modeling the spatial distribution of soil volumetric water content, provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating a method for modeling the spatial distribution of soil volumetric water content, as provided in an embodiment of the present invention. Detailed Implementation

[0029] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0030] Spatial variability structure constraint function: refers to the semivariance function model constructed based on the latitude and longitude coordinates of each station and the daily value of soil volumetric water content, which is used to characterize the spatial autocorrelation and variability of regional soil moisture. It provides the spatial constraint boundary for interpolation weight calculation to transform discrete station data into a continuous grid field.

[0031] Daily value of soil volumetric water content at grid points: refers to the value reconstructed by the spatial interpolation model on geographic grid nodes of a preset resolution, which represents the continuous distribution of shallow soil moisture on a given day. It is used to remove local sampling bias caused by uneven spatial distribution of sites.

[0032] Drought Evolution Early Warning Index: This index is a target function constructed based on historical climatological water content, combined with anomaly percentage and consecutive days of water shortage. It is used to evaluate the degree to which the current soil moisture status deviates from the normal range and to provide optimization guidance for determining the drought level.

[0033] Valid sample data: refers to the qualified data retained after the original hourly volumetric water content data of automatic soil moisture monitoring stations are screened by the lower limit threshold of 3.5%, the upper limit threshold of 52.0%, time consistency and Grubbs test algorithm, and used to participate in the daily value integration calculation at the station scale.

[0034] Soil moisture attribute feature vector: refers to a multidimensional mathematical matrix formed by spatially registering and assimilating daily station data, latitude and longitude coordinates, and elevation attributes. It is used as the input feature set for spatial variation structure modeling in the processing system.

[0035] Spatial drift component: refers to the component that causes soil moisture observations to systematically deviate from the overall regional level due to differences in the local geographical representativeness of the site selection, such as being located in a depression, at the foot of a slope, or near irrigation facilities. It is used to correct the non-stationarity of the interpolation results.

[0036] Soil moisture spatial enhancement grid set: refers to a high-fidelity gridded soil moisture product synthesized on the basis of a standard grid background field by filtering out spatial drift components and enhancing local topographic disturbance field information, which is used to improve the characterization depth of soil moisture spatial heterogeneity.

[0037] Humidity decay rate: refers to the decay slope of the intrinsic mode function component extracted from gridded soil moisture time-series data within a continuous monitoring period on the envelope line, used to quantify the rate of physical process of soil receding from wet to dry.

[0038] The coupled feature vector of spatial heterogeneity and temporal evolution of soil moisture refers to the composite feature set constructed by multidimensionally mapping the local water content differences extracted from soil layers at different depths with the rate of change of humidity decay in a spatiotemporal dimension. It is used to describe the coordinated change pattern of regional soil moisture in spatial and temporal dimensions.

[0039] Soil Moisture Anomaly Pattern Feature Database: This refers to a prior knowledge base established by clustering algorithms, which includes multiple standard soil moisture spatial distribution center vectors, based on monitoring feedback sample sets covering different landform types and soil textures, to provide a benchmark for drought assessment.

[0040] The optimal soil moisture state decision hyperplane refers to a multidimensional decision boundary that can accurately distinguish between "normal" and different levels of "drought and moisture deficiency" states, obtained in a high-dimensional geographic attribute space through kernel function mapping and training with support vector machines or hybrid heuristic search algorithms.

[0041] Environmental heterogeneity interferes with signal modulation of spatial extrapolation results: This refers to the systematic deviation in gridded reconstruction results caused by the interaction between environmental factors such as local soil texture, topographical obstruction, and micro-meteorological conditions and the boundary conditions of the interpolation algorithm.

[0042] Kernel function: refers to a mathematical method for calculating the spatial dot product of geographic attribute features under the condition of satisfying high-dimensional mapping, and is used to construct drought level determination models in nonlinear environments.

[0043] Hybrid heuristic search: refers to an iterative algorithm that combines global gradient descent with local stochastic optimization, used to search for the optimal evaluation solution of interpolation weights under the constraint of spatial heterogeneity in complex soil.

[0044] Second-order fine spatial diagnostic vector: refers to a high-order descriptive factor formed by extracting instantaneous spatial gradient flow and narrow-band directional gradient spectrum from the initial soil moisture spatial distribution grid set, and fusing the contrast factor of local abnormal mutation patches with the slope aspect distribution difference characteristics. It is used to capture the real local moisture variation signal caused by topographic convergence lines or irrigation facility leakage.

[0045] Dynamic weight compensation value: refers to the compensation amount automatically generated based on the reverse tracing results of the grid point anomaly fluctuation time series characteristics during continuous monitoring period, according to the trend decay of interpolation accuracy, and is used to correct the semivariance function parameters inside the spatial variation structure constraint function, in order to maintain the long-term stability of the gridded model.

[0046] like Figures 1-2 As shown in some embodiments of this application, this embodiment provides a method for modeling the spatial distribution of soil volumetric water content, including: Step S100: Obtain hourly data of shallow stratified soil volumetric water content from each automatic soil moisture monitoring station within the provincial jurisdiction. Extract effective sample data from each station based on preset lower limit threshold test, upper limit threshold test, time consistency test and Grubbs test algorithm. Calculate the average value of the hourly effective samples of the three soil layers within each station scale, and then obtain the daily value of shallow soil volumetric water content for the station on that day.

[0047] Specifically, when extracting valid sample data from each station based on a pre-set verification algorithm, the process includes: capturing the original dielectric response pulse signals of the shallow 10 cm, 20 cm, and 30 cm soil layers in real time using time-domain acquisition units deployed at each automatic water station; calculating the instantaneous soil volumetric water content of each soil layer based on the Topp formula and pulse intensity; and adjusting the sampling window length of the fusion program and the initial weights of the multi-soil layer mean integration according to the date signal. For example, the multi-soil layer fusion calculation program is triggered after the date signal is detected, i.e., after the last data collection at 23:00 on the same day is completed.

[0048] Understandably, by introducing a multi-level quality control and fusion calculation model at the station scale, real-time dynamic screening of the validity of raw monitoring data can be achieved. First, relying on the time-domain acquisition unit, dielectric pulses of each soil layer are continuously sampled. Using numerical boundary thresholds and time consistency criteria, discrete error data susceptible to instantaneous sensor fluctuations or electromagnetic interference are transformed into a valid sample set eligible for computation. Based on this, a diurnal trigger mechanism is introduced to dynamically adjust the mean calculation range according to the valid sample status of the three soil layers, enabling the daily station values ​​to adapt to the vertical moisture distribution characteristics of different soil types (such as sandy soil, loam, and clay). Furthermore, when the system scans and finds that all data for the day has passed quality control, the integration logic is automatically activated. By performing equal-weighted or weighted averaging of the valid means from the 10cm, 20cm, and 30cm layers, an objective representation of the overall shallow soil moisture status at the station is achieved, improving the accuracy of extracting true soil moisture information.

[0049] For example, in the processing of a representative site, the system extracted 22, 24, and 23 valid hourly samples from the 10 cm, 20 cm, and 30 cm soil layers, respectively. Based on this, the mean values ​​of the valid samples from the three soil layers were calculated to be 18.5 degrees, 22.1 degrees, and 25.8 degrees, respectively. By calculating the average of the three mean values ​​with equal weights, the daily shallow soil volumetric water content at this site was found to be 22.1%, providing high-confidence data input for subsequent spatial interpolation modeling.

[0050] Step S200: Collect daily station value data and corresponding latitude and longitude coordinate data of all stations in the province on the same day, and construct a spatial variation structure constraint function based on the latitude and longitude coordinate data.

[0051] Specifically, when extracting high-resolution gridded daily values ​​representing the continuous distribution across the province, the process includes: decomposing the neighborhood of the grid point to be interpolated into a standard grid background field and a local topographic disturbance field; reconstructing the processed background field and disturbance field point by point, and synthesizing a spatially enhanced gridded set of soil moisture using a Kriging weighted algorithm; and vertically slicing the spatially enhanced gridded set of soil moisture to extract the local water content difference ΔZvertical at different soil depths of 10 cm, 20 cm, and 30 cm within the spatial grid. Specifically, when collecting daily station data and corresponding latitude and longitude coordinates, the process includes: sending station geometric topology sensing signals to the gridding processing engine through spatial registration units, and receiving the station's position vector values ​​under the universal transverse Mercator projection framework. Collect daily shallow water value data for each station. and station elevation attribute data Reconstruct the effective soil moisture attribute feature vector of the site within the current spatial field. ,in The standard deviation represents the local topographical characteristics of the site.

[0052] Understandably, by spatially registering and reconstructing the attributes of daily data from various stations, a spatial variation structure constraint function is constructed, thereby providing spatial dependency constraint boundaries for gridded reconstruction. First, the spatial registration unit performs a unified geographic projection transformation on the daily values ​​of stations scattered across different geographical locations to determine if there are any spatial mismatches caused by drift in station latitude and longitude records. Simultaneously, the data assimilation unit fuses the station elevation attributes in real time, reconstructing the moisture attribute feature vectors under different elevation gradients. Further, spatial semivariogram analysis is performed on the station location vectors, daily values, and elevation attributes to construct a unified variation structure function that describes "sampling distance - moisture variability - topographic relief." This function allows for spatial dimensional constraints on interpolation weights in subsequent processing, eliminating reconstruction uncertainties caused by uneven spatial distribution of stations and improving the system's ability to reconstruct gridded fields in complex geographical environments.

[0053] Step S300: Input the daily value data of the stations into the spatial variation structure constraint function for grid-based reconstruction, remove the local sampling bias component caused by uneven spatial distribution of stations or local micro-topography, and extract the high-resolution grid-based daily value of soil volumetric water content that represents the continuous spatial distribution of the province.

[0054] Specifically, when removing local sampling bias components and extracting high-resolution gridded daily values, the process includes: collecting station spatial distribution density deviation samples under different seasonal precipitation conditions based on the monitoring station network, obtaining the background error covariance distribution vector of the current interpolation model; and using the formula...

[0055] Calculate the sample semivariance, where The station spacing vector, For spacing equal to The number of sites, integration interval: covering the entire province's spatial domain; based on the analysis of the spatial characteristic scale of the background error covariance using a statistical optimization algorithm, the search radius of the interpolation model is determined. With azimuth parameters .

[0056] Specifically, extracting high-resolution grid daily values ​​representing the continuous distribution across the province also includes: decomposing the neighborhood of the grid points to be interpolated into a standard grid background field and a local topographic disturbance field; reconstructing the processed background field and disturbance field point by point, and synthesizing a spatially enhanced grid set of soil moisture using the Kriging weighted algorithm; and vertically slicing the spatially enhanced grid set of soil moisture to extract the local differences in water content within the spatial grid at different soil depths of 10 cm, 20 cm, and 30 cm. .

[0057] Understandably, this step constructs a gridded reconstruction mechanism of "background field constraint + topographic disturbance stripping". First, a semivariogram analysis is performed on the station data using a spatial variation structure constraint function to remove covariance estimation biases caused by the spatial clustering or sparse distribution of stations from the global model. Then, through statistical optimization of the background error covariance distribution, the optimal search radius and anisotropy parameters are locked and applied to physically strip the spatial drift components caused by "local micro-topography or stations near irrigation facilities". Further, through point-by-point synthesis of the background field and the disturbance field, a spatially enhanced gridded set of different vertical soil layers is obtained at the regional level, ultimately forming a high-resolution and spatially continuous soil moisture distribution characteristic. This characteristic not only includes spatially continuous information on water abundance and scarcity but also incorporates local water content fluctuations caused by soil texture and topography, greatly improving the system's characterization accuracy for areas with low water content below 10% and high water content above 50%.

[0058] Table 1 below shows a comparison of the gridding effects before and after the algorithm reconstruction:

[0059] Step S400: Perform collaborative measurement and judgment between the grid point daily values ​​and the preset drought evolution early warning index; generate tiered early warning instruction labels based on the judgment results; when the soil water separation difference is found to exceed the warning level, output visualization rendering instructions to the color map driven by thematic map making.

[0060] Specifically, when co-measuring and judging gridded daily values ​​with drought evolution early warning indices, the following steps are taken: obtaining monitoring feedback sample sets covering different landform types and soil textures, and establishing a standard soil moisture spatial distribution center vector using clustering algorithms. A kernel-function-based decision model is constructed to calculate the geometric distance between the real-time soil moisture coupled feature vector and the optimal soil moisture state decision hyperplane. ; to measure collaboration Compared with the first preset drought level threshold (Upper limit of normal fluctuations) and the second preset drought level threshold Compare with the drought warning line.

[0061] Specifically, if Then, a second-order decomposition and identification is performed along the temporal dimension, including: extracting the instantaneous spatial gradient flow of the grid set using the sliding window mutation operator. This involves identifying local anomalous abrupt changes in the spatial domain; performing narrowband directional gradient spectrum analysis on the grid set to extract the differences in moisture distribution between the windward and leeward slopes on either side of the topographic elevation gradient; and adjusting the contrast factor of the local anomalous abrupt changes. By integrating the symmetry of slope aspect distribution differences, a second-order fine spatial diagnostic vector is constructed to determine whether there is a local soil moisture deficit caused by a period of continuous low precipitation or abnormal irrigation interception.

[0062] Understandably, through a dual-threshold mechanism and second-order decomposition identification, the system achieves refined control over the evolution of drought. When the co-measure is within the normal range, the system determines that the current soil moisture state is within the normal fluctuation range and does not need to trigger an early warning; when it is between the first and second thresholds, the system no longer simply performs a binary boundary, but distinguishes between genuine regional drought and moisture deficit signals and false moisture anomalies caused by topographic artifacts such as valley inversion effects or local precipitation interception by analyzing instantaneous spatial gradient flow and slope aspect moisture differences. This mechanism effectively solves the problem of false alarms caused by the representativeness bias of stations under complex terrain conditions in existing technologies. For example, if the moisture difference between the leeward and windward slopes is highly asymmetrical, and the contrast factor of local abrupt change patches is within the first preset range, it indicates the existence of persistent moisture maintenance formed by topographic convergence line shading, and is judged as non-drought local fluctuation; if If the soil moisture deficit exceeds the second threshold and the second-order fine-tuning diagnostic vector indicates that the contrast factor of a large area exceeds the limit, it is identified as a real soil moisture deficit caused by the accumulation of meteorological drought. Based on this, the system generates a tiered early warning instruction label and drives the color mapper to output a visualization rendering instruction, thereby maintaining the high signal-to-noise ratio and decision accuracy of the regional drought monitoring products.

[0063] Step S500: Extract the time series characteristics of grid point anomaly fluctuations in the continuous monitoring period, trace back to the source, and output the dynamic weight compensation value for the spatial interpolation model.

[0064] Specifically, when outputting dynamic weight compensation values, the process includes: extracting the phase displacement trajectory of the coupling feature vector over time within a preset continuous monitoring period. ;Establish With the parameters of the internal semivariance function of the interpolation algorithm The relationship model between Based on an adaptive feedback control algorithm, the displacement is converted into parameter correction instruction code data packets and transmitted in real time to the central processing unit of the processing system. This enables real-time compensation for systematic shifts in monitoring results across the entire region caused by sensor aging at automatic stations or changes in climate background. This step establishes a closed-loop feedback system of "self-diagnosis and self-repair." By performing trend evolution analysis on grid anomaly fluctuation deviation samples over multiple consecutive monitoring periods, the system can identify a trend of declining model accuracy (such as systematically lower measured values ​​due to a drop in battery voltage at automatic stations in a certain area). Subsequently, the potential shift of the semivariance function parameters is calculated using a reverse tracing model, and compensation values ​​are output in real time to adjust the range. or base value Online corrections enable the judgment model to adapt to changes in automatic monitoring network layout or the effects of slow climate shifts. This dynamic weight compensation mechanism avoids the cumbersome process of periodic retraining or recalibration required by traditional static interpolation models, ensuring the consistency of gridded monitoring products under the complex and ever-changing influences of farmland irrigation and meteorological drought.

[0065] To enable those skilled in the art to fully understand and implement this invention, the specific implementation principles of this invention are further supplemented below with a specific application scenario.

[0066] Step 1: Site Data Quality Control and Daily Value Integration. In the data center of a provincial agricultural meteorological monitoring platform, the front-end acquisition module of the processing system receives hourly dielectric pulse signals uploaded by over 3,000 automatic water stations across the province in real time. When the system clock reaches 0:00 AM, the diurnal boundary trigger signal is activated. The central processing unit (CPU), based on preset thresholds of 3.5% and 52.0% and a time consistency criterion, uses a Grubbs test module to eliminate outlier values ​​caused by sensor freezing or animal contact. It then calculates the effective sample mean values ​​for 10 cm, 20 cm, and 30 cm soil layers for each station, thereby deriving the daily shallow soil volumetric water content value for that station, eliminating abnormal fluctuations at the single-station data level.

[0067] Step 2: Spatial Variation Structure Construction. The system collects daily data, latitude and longitude coordinates, and altitude from all stations across the province. The spatial registration unit maps all stations to a unified projected coordinate system, and the data assimilation module generates attribute feature vectors containing daily values, altitudes, and station spatial configurations. The CPU uses these vectors to calculate the experimental semivariogram function, and through statistical analysis, optimizes the range at the current moment to be 150 kilometers, with a base value of 12.0 (unit: %2). It also identifies that the principal anisotropy axis follows the regional terrain in an east-west direction, providing physical constraints for subsequent interpolation.

[0068] Step 3: Gridded Reconstruction and Bias Removal. When generating a 5km × 5km regular grid, the CPU decomposes the neighborhood of each grid point into a standard background field and a topographic disturbance field. Using the Kriging algorithm, the system identifies and filters out local sampling bias caused by the high water-holding capacity of several stations distributed in the river valley lowlands. Through signal reconstruction, the system successfully removes the "patchy pseudo-wetness" caused by local topography from the continuously distributed grid field, improving the contrast of the arid core area by more than two times.

[0069] Step 4: Second-order drought identification. When the system calculates that the co-metrics of a contiguous area is in the critical range, second-order decomposition is initiated. The sliding window mutation operator captures anomalous patches with significant contrast in the spatial domain. Through narrowband directional gradient spectrum analysis, it is found that the patches are distributed in the rain shadow area of ​​the windward slope, and the difference in moisture between the leeward and windward slopes is highly asymmetrical, and it highly coincides with the low precipitation cycle in time. The system determines that this is a real meteorological drought deficit signal, rather than an artifact caused by irrigation activities, and therefore automatically generates a "moderate drought" warning instruction label, and instructs the color mapper to highlight it in orange on the GIS thematic map.

[0070] Step 5: Adaptive Parameter Correction. After six months of continuous operation, the system detected a systematic positive bias trend in the grid anomaly deviation trajectories of 12 stations in the Northwest region. Reverse tracing model analysis determined that scaling on sensor probes at some stations in this region caused zero-point drift in dielectric constant measurements. Therefore, an automatic compensation command was issued, lowering the sill value of the semivariance function for this region in the interpolation algorithm by 1.5 and outputting a warning signal to prompt maintenance. This process ensures the stability of the gridded monitoring system under long-term unattended dynamic environments.

[0071] As can be seen, the soil volumetric water content spatial distribution modeling method described in the above embodiments achieves real-time dynamic validity screening of the original monitoring data of the stations through a multi-soil layer quality control process. By constructing a spatial variation structure constraint function, this invention solves the problem of continuity discontinuity in regional spatial field reconstruction caused by the discrete distribution of stations, transforming the complex regional heterogeneous water distribution process into a computable standardized grid field, thereby improving the geospatial resolution of water monitoring. Secondly, by introducing anomaly percentage calculation and temporal secondary decomposition identification mechanism, it is possible to perform fine-level feature extraction for local anomalies caused by meteorological drought accumulation or soil texture differences, effectively identifying spatial expression biases caused by topographic artifacts. Through the synthesis of soil moisture spatial enhancement grid set and high-dimensional geographic attribute spatial mapping decision, this invention maintains a high signal-to-noise ratio of monitoring products under complex local environmental disturbances and background climate fluctuations, enhancing the system's response sensitivity to regional initial drought signals. Finally, through reverse tracing and dynamic weight compensation mechanisms, closed-loop optimization of interpolation model parameters is achieved, solving the lag problem of traditional single-point analysis logic in dealing with changes in automatic station networks or performance changes of monitoring instruments. This invention not only ensures the spatial continuity of provincial regional water monitoring, but also optimizes the gradient expression within the drought evolution path through spatiotemporal coupling feature analysis, eliminating systematic errors such as uneven station density and sample bias, and significantly improving the quality, reliability, and decision-making efficiency of agricultural meteorological disaster monitoring.

[0072] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0073] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0074] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0075] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method of modeling the spatial distribution of soil volumetric water content, applied to a data set, characterized in that, include: Real-time acquisition of hourly data on shallow soil volumetric water content from each automatic soil moisture monitoring station within the provincial jurisdiction; extraction of effective sample data from each station based on preset lower threshold test, upper threshold test, time consistency test and Grubbs test algorithms; calculation of the average value of hourly effective samples of the three soil layers at each station scale; and thus obtaining the daily value of shallow soil volumetric water content at each station. Collect daily station value data and corresponding latitude and longitude coordinate data from all stations in the province on the same day, and construct a spatial variation structure constraint function based on the latitude and longitude coordinate data; The daily values ​​of soil volumetric water content at the stations are input into the spatial variation structure constraint function for grid-based reconstruction. The local sampling bias components caused by uneven spatial distribution of stations or local micro-topography are removed, and the high-resolution grid-based daily values ​​of soil volumetric water content that represent the continuous spatial distribution of the province are extracted. The grid daily values ​​are compared with the preset drought evolution early warning index for a co-measurement judgment. If the co-measurement is lower than the first preset drought level threshold, the current soil moisture status is determined to be in the normal fluctuation range. If the co-measurement is higher than or equal to the first preset drought level threshold and lower than the second preset drought level threshold, the grid daily values ​​are decomposed in the time dimension to identify the local soil moisture deficit caused by the continuous low precipitation cycle or abnormal irrigation interception. Based on the judgment results, a tiered early warning instruction label is generated. When the soil water separation difference is found to exceed the warning level, a visualization rendering instruction is output to the color map driven by thematic map making. Extract the time-series characteristics of grid anomaly fluctuations over a continuous monitoring period, trace the source in reverse, and output dynamic weight compensation values ​​for the spatial interpolation model.

2. The method for modeling spatial distribution of soil volumetric water content according to claim 1, wherein, Based on a pre-defined testing algorithm, valid sample data for each site is extracted, including: Based on the time-domain acquisition units deployed at each automatic water station, the original dielectric response pulse signals of the shallow soil layers of 10 cm, 20 cm and 30 cm are captured in real time every hour, and the instantaneous value of soil volume water content of each soil layer is calculated by pulse intensity. The numerical boundaries of each soil layer per hour are collected by a threshold discrimination algorithm. The comparison between the hourly value and the lower threshold of 3.5% and the upper threshold of 52.0% is used to drive the effective sample screening unit to update the current time status marker value. When the date signal is detected, the multi-soil layer fusion calculation program is triggered to perform the shallow daily value integration calculation for that day.

3. The method for modeling spatial distribution of soil volumetric water content according to claim 2, wherein, When calculating the average of the three soil layers' hourly effective samples at each site scale, the following is included: The valid data bits after quality control are obtained using Formula 1, as shown below: Equation 1 wherein, is the first is the volume water content instantaneous value at the time instant, is the effective data bit reserved after the Grubbs test, is the time consistency threshold, the integration interval: from 00 to 23 o'clock of the day; The sampling window length and the initial weights of the multi-soil layer mean integration calculation program are adjusted based on the date and time signals.

4. The method for modeling spatial distribution of soil volumetric water content according to claim 1, wherein, The daily data collected from the stations and their corresponding latitude and longitude coordinates include: The spatial registration unit sends the site's geometric topology sensing signal to the gridding processing engine and receives the site's projected position vector value under the unified georeferenced frame. The daily shallow soil moisture data and elevation data of each station are collected by the data assimilation unit to reconstruct the effective soil moisture attribute feature vector of the station in the current spatial field.

5. The method for modeling the spatial distribution of soil volumetric water content as described in claim 1, characterized in that, When inputting daily station data into a spatial variability structure constraint function for gridded reconstruction, and removing local sampling bias components caused by uneven spatial distribution of stations or local micro-topography, the following steps are taken: Based on the monitoring network, samples of spatial distribution density deviation of stations under different seasonal precipitation conditions are collected to obtain the background error covariance distribution vector of the current interpolation model; Based on the analysis of the spatial characteristic scale of the background error covariance distribution vector using statistical optimization algorithms, the search radius and azimuth parameters of the interpolation model are determined. The neighborhood of the grid point to be interpolated is decomposed into a standard grid background field and a local terrain disturbance field. While keeping the standard grid background field unchanged, the spatial drift component defined by the difference in the representativeness of the site selection target is filtered out from the local terrain disturbance field by the Kriging weighted algorithm. The processed background field and disturbance field are reconstructed point by point to synthesize a spatially enhanced grid set of soil moisture.

6. The method for modeling the spatial distribution of soil volumetric water content as described in claim 5, characterized in that, When extracting high-resolution gridded daily values ​​of soil volumetric water content to characterize the continuous spatial distribution across the province, the following are included: Adaptive temporal mode decomposition is performed on the spatially enhanced grid set of soil moisture to obtain the intrinsic temporal function components corresponding to the transient excitation of the region by previous concentrated precipitation. Extract the periodic oscillation envelope of the intrinsic time series function components and calculate the rate of change of soil moisture decay during the continuous time series. Vertically slice the spatially enhanced grid set of soil moisture to extract the local moisture content difference at different soil depths within the spatial grid. By performing a spatiotemporal multidimensional mapping between the rate of change in humidity decay and the local difference in water content, a coupled feature vector of soil moisture spatial heterogeneity and temporal evolution is constructed.

7. The method for modeling the spatial distribution of soil volumetric water content as described in claim 1, characterized in that, When performing a collaborative measurement and judgment between gridded daily values ​​and a preset drought evolution early warning index, the following is included: Obtain monitoring feedback sample sets covering different landform types and soil textures, establish standard soil moisture spatial distribution center vectors based on clustering algorithms, and form a soil moisture anomaly pattern feature library; Construct a decision model based on geographic weighted regression or hybrid heuristic search, and pre-train it using the standard soil moisture spatial distribution center vector; The real-time acquired soil moisture coupled feature vector is input into a preset kernel function and mapped to a high-dimensional geographic attribute space to construct the optimal soil moisture state decision hyperplane. Calculate the geometric distance of the soil moisture coupled feature vector relative to the decision boundary of each drought level.

8. The method for modeling the spatial distribution of soil volumetric water content as described in claim 1, characterized in that, When performing a second time-series decomposition on the daily grid values ​​to identify local soil moisture deficit patterns caused by periods of sustained low precipitation or abnormal irrigation interception, the following are included: Instantaneous spatial gradient flow of soil moisture spatially enhanced grid set is extracted based on sliding window mutation operator to identify local anomalous mutation patches in spatial domain; Narrow-band directional gradient spectrum analysis was performed on the spatially enhanced grid set of soil moisture to extract the differences in moisture distribution between the windward and leeward slopes on both sides of the topographic elevation gradient. By fusing the contrast factor of local abnormal mutation patches with the slope aspect distribution difference characteristics, a second-order fine spatial diagnostic vector is constructed.

9. The method for modeling the spatial distribution of soil volumetric water content as described in claim 8, characterized in that, When fusing the contrast factor of local anomalous mutation patches with the slope aspect distribution difference characteristics to construct a second-order refined spatial diagnostic vector, the following is included: The extracted contrast factor and slope aspect distribution difference features are normalized and fused to establish a second-order fine spatial diagnostic vector that maps the soil water separation difference situation, where: When the contrast factor is in the first preset range and the slope aspect difference is asymmetrically distributed, it is determined that there is a real local water variation signal caused by topographic convergence line or irrigation facility leakage, so as to establish a second-order fine spatial diagnostic vector that maps the soil water separation difference situation. When the contrast factor exceeds the upper limit of the first preset interval, it is determined that there is a local soil moisture deficit caused by a period of continuous low precipitation. Based on the excess amplitude of the contrast factor, the logic of generating a tiered early warning instruction label is triggered to establish a second-order fine spatial diagnostic vector that maps the soil moisture deficit situation.

10. The method for modeling the spatial distribution of soil volumetric water content according to claim 1, characterized in that, When extracting the temporal characteristics of grid anomaly fluctuations over a continuous monitoring period, tracing back to the source, and outputting dynamic weight compensation values ​​for the spatial interpolation model, the following steps are taken: Trend evolution analysis is performed on grid point anomaly fluctuation deviation samples within a preset continuous monitoring period to extract the phase displacement trajectory of the feature vector over time. Establish a correlation model between the phase displacement trajectory and the parameters of the semivariance function inside the interpolation algorithm; The adaptive feedback control algorithm converts the displacement of the phase displacement trajectory into parameter correction instruction code data packets, which are then transmitted to the central processing unit in real time.

Images