Soil Moisture Retrieval Method and System Based on Microwave Scattering Mechanism and DNN Model

Abstract

This invention relates to the field of artificial intelligence and proposes a method and system for soil moisture retrieval based on microwave scattering mechanism and DNN model. The method includes: acquiring relevant field data of the planting area; preprocessing the data to extract the bipolar crop index; querying corresponding surface crop parameters to determine the growth period of different plants; performing backscattering simulation of the growth period based on a preset coupled microwave scattering model; obtaining backscattering simulation parameters and calculating actual observed values; querying the time observation sequence; constructing a multi-source time-series feature dataset in conjunction with the preset DNN model; extracting multi-source feature data; analyzing plant location information; retrieving soil moisture content; identifying content distribution characteristics; and finally generating a soil moisture retrieval report for the planting area. This invention can improve the effectiveness of soil moisture retrieval.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence, in particular to a soil moisture inversion method and system based on a microwave scattering mechanism and a DNN model. BACKGROUND

[0002] In the face of many challenges in global ecological environment and agricultural development today, accurate determination of soil moisture content has become a key factor in protecting ecological balance and promoting sustainable agricultural development. With the rapid progress of science and technology, especially under the background of the continuous development of microwave remote sensing technology and artificial intelligence technology, there is a higher requirement for the accuracy and efficiency of soil moisture inversion.

[0003] Currently, traditional soil moisture inversion methods mainly rely on field sampling and laboratory analysis. This method is not only time-consuming and labor-intensive, but also difficult to achieve real-time monitoring on a large scale. In addition, although some inversion methods based on microwave radiation transmission models have certain feasibility in theory, the uncertainty and complexity of model parameters affect the effect in practical application. Therefore, a soil moisture inversion method based on a microwave scattering mechanism and a DNN model is needed to improve the effect of soil moisture inversion. SUMMARY

[0004] The present application provides a soil moisture inversion method and system based on a microwave scattering mechanism and a DNN model, which aims to improve the effect of soil moisture inversion.

[0005] Obtain the corresponding field-related data of the to-be-planted area, which includes dual-polarized satellite SAR data, multi-spectral satellite remote sensing data and field observation data;

[0006] Preprocess the field-related data to obtain preprocessed field data, extract the dual-polarized crop index in the preprocessed field data, query the corresponding ground crop parameters of the dual-polarized crop index, and determine the growth period of different plants in the to-be-planted area based on the ground crop parameters;

[0007] Based on the preset microwave scattering mechanism model, the growth period of different plants in the to-be-planted area is simulated by backscattering to obtain backscattering simulation parameters, and the actual observation value corresponding to the backscattering simulation parameters is calculated;

[0008] Query the time observation sequence in the actual observation value, combine the preset DNN model with the time observation sequence, construct the multi-source time sequence feature data set corresponding to the to-be-planted area, and extract the multi-source feature data in the multi-source time sequence feature data set;

[0009] Based on the multi-source feature data, position information of the plants to be planted in the to-be-planted area is analyzed, based on the position information, soil moisture content corresponding to the to-be-planted area is inversed, a content distribution characteristic corresponding to the soil moisture content is identified, and based on the content distribution characteristic, an inversion report of the soil moisture of the to-be-planted area is generated.

[0010] Optionally, the extracting the dual-polarization crop index in the pre-processed field data comprises:

[0011] The pre-processed field data is subjected to data unit segmentation to obtain data segmentation units;

[0012] The horizontal polarization data and the vertical polarization data in the data segmentation units are extracted;

[0013] The horizontal polarization data and the vertical polarization data are subjected to filtering processing to obtain horizontal polarization filtered data and vertical polarization filtered data;

[0014] Based on the horizontal polarization filtered data and the vertical polarization filtered data, a dual-polarization ratio corresponding to the pre-processed field data is calculated;

[0015] Based on the dual-polarization ratio, the dual-polarization crop index in the pre-processed field data is determined.

[0016] Optionally, the calculating the dual-polarization ratio corresponding to the pre-processed field data based on the horizontal polarization filtered data and the vertical polarization filtered data comprises:

[0017] The dual-polarization ratio corresponding to the pre-processed field data is calculated.

[0018] Optionally, the determining the growth period corresponding to different plants to be planted in the to-be-planted area based on the surface crop parameter comprises:

[0019] The surface crop parameter is subjected to parameter screening to obtain a crop screening parameter;

[0020] A crop feature corresponding to the crop screening parameter is identified, and a crop feature index corresponding to the crop feature is queried;

[0021] Based on the crop feature index, a period stage corresponding to the surface crop parameter is analyzed;

[0022] Based on the period stage, the growth period corresponding to different plants to be planted in the to-be-planted area is determined.

[0023] Optionally, the performing backscattering simulation on the growth period corresponding to different plants to be planted in the to-be-planted area based on the preset microwave scattering mechanism model comprises:

[0024] Collect detailed parameters of the growth period of different plant species in the planting area;

[0025] Based on the detailed parameters, the parameters of the preset microwave scattering mechanism model are set;

[0026] And using the coupled microwave scattering model with pre-set parameters, the crop framework corresponding to the planting area is constructed;

[0027] Identify the crop framework parameters corresponding to the crop framework;

[0028] Backscattering simulation was performed on the crop framework parameters to obtain the backscattering simulation parameters.

[0029] Optionally, querying the time observation sequence in the actual observations includes:

[0030] Collect the complete dataset corresponding to the actual observations, and extract the observation fields from the complete dataset;

[0031] The observed fields are standardized to obtain standardized fields;

[0032] Mark the field index corresponding to the standardized field, and set the time interval corresponding to the field index;

[0033] Based on the time interval, query the time observation sequence in the actual observation value.

[0034] Optionally, the step of combining a preset DNN model with the time observation sequence to construct a multi-source time-series feature dataset corresponding to the planting area includes:

[0035] Identify the data adaptation format in the preset DNN model;

[0036] Based on the adapted format, adjust the sequence observation data corresponding to the time observation sequence;

[0037] Extract multi-source sequence features from the sequence observation data;

[0038] The multi-source sequence features are transformed into vectors to obtain multi-source feature vectors;

[0039] The multi-source feature vectors are processed using a preset DNN model to obtain the multi-source time-series feature dataset corresponding to the planting area.

[0040] Optionally, the step of retrieving the soil moisture content corresponding to the planting area based on the location information includes:

[0041] Query the geographic coordinates in the location information;

[0042] Based on the geographical coordinates, analyze the land use type corresponding to the area to be planted;

[0043] Based on the land use type, determine the coverage area corresponding to the area to be planted;

[0044] Based on the coverage area, the soil moisture content in the planting area is analyzed;

[0045] Based on the soil moisture content ratio, the soil moisture content corresponding to the planting area is calculated.

[0046] To address the aforementioned problems, this invention also provides a soil moisture retrieval system based on microwave scattering mechanism and DNN model, the system comprising:

[0047] The field data module is used to acquire relevant field data corresponding to the planting area. The relevant field data includes: dual-polarization spaceborne SAR data, multispectral satellite remote sensing data, and field observation data.

[0048] The crop parameter module is used to preprocess the relevant field data to obtain preprocessed field data, extract the bipolar crop index from the preprocessed field data, query the surface crop parameters corresponding to the bipolar crop index, and determine the growth period of different plant species in the planting area based on the surface crop parameters.

[0049] The observation value calculation module is used to perform backscattering simulation on the growth period of different plant species in the planting area based on a preset microwave scattering mechanism model, obtain backscattering simulation parameters, and calculate the actual observation values ​​corresponding to the backscattering simulation parameters.

[0050] The feature extraction module is used to query the time observation sequence in the actual observation values, combine the preset DNN model with the time observation sequence, construct the multi-source time series feature dataset corresponding to the planting area, and extract the multi-source feature data in the multi-source time series feature dataset.

[0051] The report generation module is used to analyze the location information of the plants in the planting area based on the multi-source feature data, invert the soil moisture content corresponding to the planting area based on the location information, identify the content distribution characteristics corresponding to the soil moisture content, and generate an inversion report of soil moisture in the planting area based on the content distribution characteristics.

[0052] First, by acquiring relevant field data corresponding to the planting area, this invention can comprehensively and accurately understand the actual conditions of the planting area, providing rich and reliable basic information for soil moisture inversion and providing a basis for subsequent analysis of the relationship between soil moisture and vegetation growth, thereby improving the accuracy of the inversion results. Simultaneously, by preprocessing the relevant field data, this invention obtains preprocessed field data, which can remove noise and outliers, improving data quality and accuracy. It can also unify and standardize data from different sources and formats, facilitating data integration and comprehensive utilization. Based on a preset microwave scattering mechanism model, this invention performs backscattering simulations on the growth stages of different plant species in the planting area, obtaining backscattering simulation parameters, which helps to deeply understand the electromagnetic scattering characteristics of crops at different growth stages. This invention can predict the scattering patterns of microwaves by plants at different growth stages in advance, providing a theoretical basis for subsequent remote sensing monitoring and data analysis. This allows for more accurate interpretation of information in remote sensing images and improves the precision of crop monitoring. By querying the time observation sequence in the actual observation values, this invention helps to explore the influence mechanism of environmental factors and plant growth stages on backscattering, revealing patterns related to plant growth cycles, meteorological cycles, or other periodic phenomena. This provides a strong basis for prediction and model optimization. Furthermore, based on the multi-source feature data, this invention analyzes the location information of plants in the planting area, enabling precise planning of the planting area. Understanding the location distribution of plants allows for a more rational arrangement of the planting layout, avoiding overcrowding or sparseness between plants, thereby optimizing land use efficiency and increasing yield per unit area. Therefore, the soil moisture inversion method and system based on microwave scattering mechanism and DNN model proposed in this invention can improve the effect of soil moisture inversion. Attached Figure Description

[0053] Figure 1 This is a schematic flowchart of a soil moisture inversion method based on microwave scattering mechanism and DNN model provided in an embodiment of the present invention.

[0054] Figure 2 This is a schematic diagram of a soil moisture inversion system based on microwave scattering mechanism and DNN model provided in an embodiment of the present invention.

[0055] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0056] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0057] This application provides a soil moisture inversion method based on microwave scattering mechanism and DNN model. The executing entity of the soil moisture inversion method based on microwave scattering mechanism and DNN model includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application embodiment: a server, a terminal, etc. In other words, the soil moisture inversion method based on microwave scattering mechanism and DNN model can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0058] Reference Figure 1 The diagram shown is a flowchart illustrating a soil moisture retrieval method based on microwave scattering mechanism and DNN model according to an embodiment of the present invention. In this embodiment, the soil moisture retrieval method based on microwave scattering mechanism and DNN model includes:

[0059] S1. Obtain relevant field data corresponding to the planting area. The relevant field data includes: dual-polarization spaceborne SAR data, multispectral satellite remote sensing data, and field observation data.

[0060] This invention, by acquiring relevant field data corresponding to the planting area, can comprehensively and accurately understand the actual condition of the planting area, providing rich and reliable basic information for soil moisture inversion, and providing a basis for subsequent analysis of the relationship between soil moisture and vegetation growth, thereby improving the accuracy of the inversion results.

[0061] The aforementioned field-related data refers to a series of data closely related to the actual situation of the area to be planted. Specifically, it includes dual-polarization spaceborne SAR data, which reflects the characteristics of the soil surface; multispectral satellite remote sensing data, which can obtain information related to vegetation; and field observation data, which is real and reliable data obtained directly in the planting area through actual measurement and observation. Optionally, the acquisition of the field-related data corresponding to the planting area can be achieved through satellite remote sensing technology, such as the Sentinel series satellites, which can provide high-quality SAR data and multispectral data.

[0062] In detail, the relevant field data includes: dual-polarization spaceborne SAR data, multispectral satellite remote sensing data, and field observation data. The dual-polarization spaceborne SAR data refers to data obtained by observing the Earth's surface using a spaceborne synthetic aperture radar (SAR) system with both horizontal (H) and vertical (V) polarization. The multispectral satellite remote sensing data refers to data obtained by imaging the Earth's surface in multiple different spectral bands using multispectral sensors onboard satellites. Different spectral bands can reflect the reflection and radiation characteristics of ground objects in different wavelength ranges, thus distinguishing different types of ground objects, such as vegetation, water bodies, and soil, and obtaining information such as vegetation growth status and land use. The field observation data refers to data directly obtained in the planting area through on-site measurement, sampling, and surveys. This data includes, but is not limited to, information such as soil texture, soil temperature, soil moisture, vegetation height, vegetation density, and crop growth stage. Field observation data is characterized by high precision and high reliability, providing direct reference for soil moisture retrieval. It can also be used to verify and calibrate the results of other remote sensing data and models.

[0063] S2. Perform data preprocessing on the relevant field data to obtain preprocessed field data, extract the bipolar crop index from the preprocessed field data, query the surface crop parameters corresponding to the bipolar crop index, and determine the growth period of different plant species in the planting area based on the surface crop parameters.

[0064] This invention obtains preprocessed field data by preprocessing the relevant field data. This process can remove noise and outliers from the data, improve the quality and accuracy of the data, and unify and standardize data from different sources and in different formats, making it easier to integrate and utilize the data.

[0065] The preprocessed field data refers to the data obtained after performing a series of processing operations on the acquired field-related data. These processing operations may include, but are not limited to, data cleaning (removing noise, outliers, etc.), data calibration (correcting sensor errors, etc.), data format conversion (unifying to a format that is easy for subsequent processing), data fusion (integrating data from different sources), and data dimensionality reduction (reducing data dimensions to improve processing efficiency). Optionally, the data preprocessing of the field-related data can be achieved through data preprocessing tools, such as Scikit-learn, Pandas, etc.

[0066] Furthermore, by extracting the bipolar crop index from the preprocessed field data, this invention can provide a deeper understanding of crop growth conditions, such as crop health, water status, and changes in growth stages. This helps farmers and agricultural managers to take timely and appropriate measures, such as precision irrigation, fertilization, and pest and disease control, thereby improving crop yield and quality.

[0067] The bipolar crop index is an index used to describe crop characteristics, determined based on the bipolar ratio and other relevant information. It can comprehensively reflect information such as crop growth status, water content, and vegetation coverage, and is of great significance for agricultural production management and ecological environment monitoring.

[0068] As an embodiment of the present invention, the extraction of the dual-polarization crop index from the preprocessed field data includes: segmenting the preprocessed field data into data units to obtain data segmentation units; extracting horizontally polarized data and vertically polarized data from the data segmentation units; filtering the horizontally polarized data and the vertically polarized data to obtain horizontally polarized filtered data and vertically polarized filtered data; calculating the dual-polarization ratio corresponding to the preprocessed field data based on the horizontally polarized filtered data and the vertically polarized filtered data; and determining the dual-polarization crop index in the preprocessed field data based on the dual-polarization ratio.

[0069] The data segmentation unit refers to a smaller set of data into which preprocessed field data is divided according to certain rules (such as time intervals and geographic coordinates). Each data segmentation unit contains data within a specific time period and region, allowing for more detailed data analysis and processing. The horizontal polarization data refers to the field observation data corresponding to the horizontal vibration of the electric field vector during electromagnetic wave propagation, reflecting the scattering and reflection characteristics of ground objects under horizontal polarization. The vertical polarization data refers to the field observation data corresponding to the vertical vibration of the electric field vector. Similar to the horizontal polarization data, the vertical polarization data also includes the electromagnetic characteristics of ground objects under vertical polarization. The information is as follows: The horizontal polarization filtered data refers to the data obtained after filtering the horizontal polarization data. The purpose of filtering is to remove noise, interference, and other irrelevant information from the horizontal polarization data, and to retain or highlight useful signals related to the dual polarization crop index, thereby improving the quality and reliability of the data. The vertical polarization filtered data refers to the data obtained after filtering the vertical polarization data. It is used to remove noise and interference from the vertical polarization data and to provide more accurate information for subsequent calculations. The dual polarization ratio is a ratio calculated from the horizontal polarization filtered data and the vertical polarization filtered data. It characterizes certain characteristics of ground features by comparing the differences between data under horizontal polarization and vertical polarization modes.

[0070] Furthermore, the data unit segmentation of the preprocessed field data can be achieved based on time intervals and geographic grid division. For example, the data can be divided into different time units according to fixed time intervals (such as daily, weekly, etc.), and the study area can be divided into regular grids according to geographic coordinates, with each grid serving as a geographic unit. This yields data segmentation units determined by time and geographic location. The extraction of horizontal and vertical polarization data from the data segmentation units can be achieved using the ENVI tool. For example, the preprocessed field data file can be read using ENVI, and the corresponding horizontal and vertical polarization band data can be extracted using specific band selection functions. The filtering of the horizontal and vertical polarization data can be achieved using filtering algorithms, such as median filtering, mean filtering, and Gaussian filtering. The calculation of the bipolarization ratio corresponding to the preprocessed field data can be achieved using the following calculation formula. The determination of the bipolarization crop index in the preprocessed field data can be achieved using machine learning algorithms, such as support vector machines and random forests.

[0071] As an embodiment of the present invention, the step of calculating the dual polarization ratio corresponding to the preprocessed field data based on the horizontal polarization filtered data and the vertical polarization filtered data includes:

[0072] The dual polarization ratio corresponding to the preprocessed field data is calculated using the following formula:

[0073]

[0074] Wherein, DPR represents the dual polarization ratio corresponding to the preprocessed field data, and H... filtered V represents the data value corresponding to the horizontal polarization filtered data, Ha represents the weight coefficient corresponding to the horizontal polarization filtered data, Hb represents the offset corresponding to the horizontal polarization filtered data, Hc represents the constant term corresponding to the horizontal polarization filtered data, and V represents the data value corresponding to the horizontal polarization filtered data. filtered Vd represents the data value corresponding to the vertical polarization filtered data, Ve represents the weight coefficient corresponding to the vertical polarization filtered data, Vf represents the offset corresponding to the vertical polarization filtered data, and Vf represents the constant term corresponding to the vertical polarization filtered data.

[0075] In detail, the weighting coefficients corresponding to the horizontal polarization filter data refer to the factors used to measure the relative importance of the horizontal polarization filter data in the process of calculating the dual polarization ratio, which determine the degree of influence of the horizontal polarization filter data on the final result; the offset corresponding to the horizontal polarization filter data refers to the factor used to avoid invalid or unstable results in logarithmic operations. In actual horizontal polarization filter data, there may be some data points close to zero or negative values. If logarithmic operations are directly performed on these data, the calculation results will be meaningless or produce large errors. By adding an appropriate offset, all data values ​​can be adjusted to the positive range, thereby ensuring the feasibility of logarithmic operations; the constant term corresponding to the horizontal polarization filter data refers to the term that plays a fine-tuning role in the formula, which can... The horizontal polarization filter data, after logarithmic transformation and weight adjustment, is further corrected to make the calculated dual polarization ratio more consistent with the actual situation or expected results. The weight coefficients corresponding to the vertical polarization filter data are similar to those of the horizontal polarization filter data, and the weight coefficient Vd of the vertical polarization filter data is used to determine the relative importance of the vertical polarization filter data in calculating the dual polarization ratio. The offset of the vertical polarization filter data serves the same purpose as the offset of the horizontal polarization filter data, ensuring that the input value is positive in the logarithmic operation. The constant term of the vertical polarization filter data also functions similarly to the constant term of the horizontal polarization filter data, used for final fine-tuning of the vertical polarization filter data after logarithmic transformation and weight adjustment.

[0076] This invention, by querying the surface crop parameters corresponding to the bipolar crop index, can provide key information for precision agriculture, which helps to conduct in-depth research on crop growth patterns and soil-crop interaction mechanisms, and can reveal the physiological responses and adaptability of crops under different growth conditions.

[0077] The surface crop parameters refer to a series of quantitative indicators used to describe the characteristics and status of surface crops, including crop type, growth stage, vegetation coverage, etc. Optionally, the query of the surface crop parameters corresponding to the bipolar crop index can be achieved by regression models, such as linear regression, multinomial regression, nonlinear regression, etc.

[0078] Based on the aforementioned surface crop parameters, this invention determines the corresponding growth period of different plant species in the planting area, which helps to accurately grasp the key stages of crop growth. It can rationally arrange the timing and amount of irrigation and fertilization, avoid resource waste and the negative environmental impact of excessive fertilization, and at the same time ensure that crops receive sufficient nutrient supply during key growth stages, promote healthy crop growth, and improve yield and quality.

[0079] The growth period refers to the entire time from the sowing or emergence of a crop, through a series of growth and development stages, until the crop matures and is harvested. This time period is further subdivided into several different growth stages, such as the sowing period, seedling stage, tillering stage, heading stage, grain-filling stage, and maturity stage. Each stage has specific growth tasks and specific requirements for external environmental conditions.

[0080] As an embodiment of the present invention, determining the growth period of different plant species in the planting area based on the surface crop parameters includes: screening the surface crop parameters to obtain crop screening parameters; identifying the crop characteristics corresponding to the crop screening parameters; querying the crop characteristic index corresponding to the crop characteristics; analyzing the cycle stage corresponding to the surface crop parameters based on the crop characteristic index; and determining the growth period of different plant species in the planting area based on the cycle stage.

[0081] The crop screening parameters refer to the key parameters that are closely related to the crop's growth period and have significant indicative role, selected from numerous surface crop parameters through scientific analysis and judgment. For example, leaf area growth rate, stem diameter variation, and chlorophyll content may be selected as key crop screening parameters. The crop characteristics refer to the unique and observable properties and features exhibited by the crop during its growth process. These can be morphological, such as plant shape, leaf color, and texture, or physiological, such as photosynthetic efficiency and respiration intensity. The crop characteristic index refers to standardized numerical indicators set to more accurately describe and quantify crop characteristics. For example, an index calculated using a specific formula to represent the degree of leaf senescence or an index representing root vitality. The cycle stage refers to a specific time period in the crop's growth process, during which the crop's growth and development exhibit relatively stable and regular characteristics, such as the vigorous period of vegetative growth or the critical period for reproductive organ formation.

[0082] Furthermore, the parameter screening of the surface crop parameters can be achieved through parameter screening methods, such as correlation analysis and principal component analysis; the identification of crop characteristics corresponding to the crop screening parameters can be achieved through the Python ScikitLearn library, such as using clustering algorithms in the Python ScikitLearn library for analysis; the querying of crop characteristic indices corresponding to the crop characteristics can be achieved through R language, such as using R language for statistical analysis and calculating crop characteristic indices; the analysis of the periodic stages corresponding to the surface crop parameters can be achieved through stage analysis methods, such as comparative analysis and fuzzy logic reasoning; the determination of the growth period corresponding to different plant species in the planting area can be achieved through reasoning methods, such as comparative analysis and fuzzy logic reasoning.

[0083] S3. Based on the preset microwave scattering mechanism model, backscattering simulation is performed on the growth period of different plant species in the planting area to obtain backscattering simulation parameters, and the actual observed values ​​corresponding to the backscattering simulation parameters are calculated.

[0084] This invention, based on a pre-defined microwave scattering mechanism model, simulates backscattering at different growth stages of different plant species in the planting area to obtain backscattering simulation parameters. This helps to gain a deeper understanding of the electromagnetic scattering characteristics of crops at different growth stages, and can predict the scattering patterns of microwaves by plants at different growth stages in advance. This provides a theoretical basis for subsequent remote sensing monitoring and data analysis, thereby more accurately interpreting the information in remote sensing images and improving the accuracy of crop monitoring.

[0085] The preset microwave scattering mechanism model refers to a pre-established mathematical model used to describe and calculate the scattering of electromagnetic waves during their interaction with plants and the surrounding environment. The backscattering simulation parameters refer to the parameters obtained by performing backscattering simulation calculations on the constructed crop framework. These parameters quantitatively describe the intensity, frequency characteristics, polarization characteristics, and other information of electromagnetic waves backscattered after interacting with the crop, and are used to analyze and evaluate the crop's growth status, structural characteristics, and interaction with electromagnetic waves.

[0086] As an embodiment of the present invention, the backscattering simulation of the growth period of different plant species in the planting area based on the preset microwave scattering mechanism model to obtain backscattering simulation parameters includes: collecting detailed parameters of the growth period of different plant species in the planting area; setting parameters of the preset microwave scattering mechanism model based on the detailed parameters; constructing a crop framework corresponding to the planting area using the parameter-set coupled microwave scattering model; identifying the crop framework parameters corresponding to the crop framework; and performing backscattering simulation on the crop framework parameters to obtain backscattering simulation parameters.

[0087] The detailed parameters refer to the data on various specific characteristics and attributes of different plant species in the planting area during their corresponding growth stages, including but not limited to plant height, stem diameter, leaf area, leaf shape, water content, plant density, and soil moisture, roughness, and other related information. The crop framework refers to a virtual model constructed in a preset microwave scattering mechanism model based on the collected detailed parameters, which can represent the growth and structural characteristics of plants in the planting area. It includes information such as plant morphology, structure, and relationship with the surrounding environment. The crop framework parameters refer to various specific quantitative indicators describing the crop framework, such as the geometric dimensions of the framework, the proportion and distribution of its components, and material properties (such as dielectric constant). These parameters are used to accurately characterize the features of the crop framework.

[0088] Furthermore, the detailed parameters of the growth periods of different plant species in the planting area can be collected using remote sensing tools, such as ENVI; the parameter settings of the preset microwave scattering mechanism model can be achieved using parameter import tools, such as ETL and Web; the construction of the crop framework corresponding to the planting area can be achieved using the coupled microwave scattering model with pre-set parameters; the identification of the crop framework parameters corresponding to the crop framework can be achieved using parameter identification tools, such as SolidWorks; and the backscattering simulation of the crop framework parameters can be achieved using simulation methods, such as numerical simulation and Monte Carlo simulation.

[0089] This invention, by calculating the actual observed values ​​corresponding to the backscattering simulation parameters, can accurately monitor the growth status, health level, and yield prediction of crops, providing a scientific basis for decision-making in agricultural production and realizing the efficient utilization and precise management of agricultural resources.

[0090] The actual observed values ​​refer to real data on backscattering obtained through actual measurement, monitoring or observation methods, which reflect the actual backscattering situation generated after electromagnetic waves interact with objects in the real environment.

[0091] As an embodiment of the present invention, the calculation of the actual observed values ​​corresponding to the backscattering simulation parameters includes:

[0092]

[0093] Where Obv represents the actual observed value corresponding to the backscattering simulation parameters, N represents the number of observations corresponding to the backscattering simulation parameters, i represents the observation index corresponding to the backscattering simulation parameters, and f min f represents the lower limit of frequency. max S represents the upper limit of frequency. sim (f,θ i ,φ i ) indicates that at frequency f and incident angle θ i Azimuth φ i The backscattering simulation value under the following conditions, S obs (f,θ i ,φ i () represents the frequency f and the incident angle θ i Azimuth φ i The corresponding actual observed backscattering value, λ represents the environmental influence weighting coefficient in the backscattering simulation parameters, M env (i) represents the environmental parameter matrix at the i-th observation.

[0094] Specifically, the lower frequency limit and the upper frequency limit refer to the minimum value of the considered frequency range, while the upper frequency limit is the maximum value; the simulated backscattering value refers to the electromagnetic wave scattering value in the backscattering direction predicted by theoretical models and calculation methods; the actual observed backscattering value refers to the electromagnetic wave scattering data in the backscattering direction directly measured in a real scene using specialized instruments and equipment; the environmental impact weighting coefficient is a value used to quantify the degree of influence of environmental factors on backscattering, which determines the proportion of environmental parameters in the calculation of actual observed values. A larger weighting coefficient means that the environmental factors have a greater impact on the results, and vice versa; the environmental parameter matrix refers to a multidimensional array or data table used to describe environmental characteristics, which can contain quantitative values ​​of various environmental factors such as soil moisture, temperature, wind speed, and atmospheric conditions.

[0095] S4. Query the time observation sequence in the actual observation values, combine the preset DNN model with the time observation sequence to construct the multi-source time series feature dataset corresponding to the planting area, and extract the multi-source feature data in the multi-source time series feature dataset.

[0096] This invention, by querying the time observation sequence in the actual observation values, helps to explore in depth the influence mechanism of environmental factors and plant growth stages on backscattering, revealing the patterns related to plant growth cycles, meteorological cycles or other periodic phenomena, and providing a strong basis for prediction and model optimization.

[0097] The time observation sequence refers to a series of observations arranged in chronological order, reflecting the changes in the state or characteristics of the observed object at different points in time.

[0098] As an embodiment of the present invention, querying the time observation sequence in the actual observations includes: collecting the complete dataset corresponding to the actual observations; extracting the observation fields from the complete dataset; standardizing the observation fields to obtain standardized fields; marking the field index corresponding to the standardized fields; setting the time interval corresponding to the field index; and querying the time observation sequence in the actual observations based on the time interval.

[0099] The complete dataset refers to a collection of all raw data related to actual observations, covering various observation conditions, environmental factors, measurement parameters, and other information. The observation field refers to a specific data column or data area in the complete dataset specifically used to record observation results or measurement values. The standardized field refers to an observation field that has undergone a series of standardized processing and transformations to achieve uniformity and standardization in format, units, and precision. The field index refers to a unique identifier or number assigned to each field (including standardized fields) in the dataset. The time interval refers to a fixed time period set on the time axis for grouping, aggregating, or analyzing data, such as hourly, daily, or weekly intervals.

[0100] Furthermore, the collection of the complete dataset corresponding to the actual observations can be achieved using database management tools, such as MySQL; the extraction of observation fields from the complete dataset can be achieved using field extraction methods, such as data filtering and field selection; the standardization of the observation fields can be achieved using standardization tools, such as using functions in the NumPy library; the marking of the field indexes corresponding to the standardized fields can be achieved using index marking tools, such as MySQL and Python; the setting of the time interval corresponding to the field indexes can be achieved using SPSS tools, such as using SPSS to analyze the periodicity of data to determine the interval; and the querying of the time observation sequence in the actual observations can be achieved using sequence query methods, such as time sorting query and sliding window query.

[0101] This invention constructs a multi-source time-series feature dataset corresponding to the planting area by combining a preset DNN model with the time observation sequence. This dataset can uncover the complex patterns and potential relationships hidden in the time observation sequence, thereby more accurately depicting the feature changes of the planting area.

[0102] The pre-set DNN model refers to a deep neural network model that is designed and trained in advance according to specific tasks and data characteristics. It has a specific structure and parameters and is used to process and analyze input data to obtain useful information or make predictions. The multi-source time series feature dataset refers to a collection of multi-source time series features that have been processed and extracted. It is organized according to certain structures and rules and is used for subsequent analysis and application.

[0103] As an embodiment of the present invention, the step of constructing a multi-source time-series feature dataset corresponding to the planting area by combining a preset DNN model with the time observation sequence includes: identifying the adaptation format of the data in the preset DNN model; adjusting the sequence observation data corresponding to the time observation sequence based on the adaptation format; extracting multi-source sequence features from the sequence observation data; performing vector transformation on the multi-source sequence features to obtain multi-source feature vectors; and performing time-series processing on the multi-source feature vectors using the preset DNN model to obtain the multi-source time-series feature dataset corresponding to the planting area.

[0104] The adaptation format refers to the specific organizational form, data type, dimension, and arrangement of data that the preset DNN model can accept and process; the sequence observation data refers to the data composed of observation values ​​recorded in chronological order, reflecting the state or characteristics of the observed object at different time points; the multi-source sequence features refer to the significant characteristics or patterns extracted from the sequence observation data that come from multiple different sources (such as meteorology, soil, crops themselves, etc.) and have time-series properties; the multi-source feature vector refers to the mathematical vector obtained after numericalizing and vectorizing the extracted multi-source sequence features.

[0105] Furthermore, the adaptation format of the data in the preset DNN model can be achieved through model analysis software, such as Netron; the adjustment of the sequence observation data corresponding to the time observation sequence can be achieved through data adjustment methods, such as data resampling and data scaling; the extraction of multi-source sequence features from the sequence observation data can be achieved through feature extraction methods, such as sliding window analysis and Fourier transform; and the vector transformation of the multi-source sequence features can be achieved through vector transformation methods, such as One-Hot encoding and numerical vector methods.

[0106] S5. Based on the multi-source feature data, analyze the location information of the plants in the planting area, and based on the location information, invert the soil moisture content corresponding to the planting area, identify the content distribution characteristics corresponding to the soil moisture content, and generate an inversion report of soil moisture in the planting area based on the content distribution characteristics.

[0107] Based on the multi-source feature data, this invention analyzes the location information of the plants in the planting area, enabling precise planning of the planting area, understanding the location distribution of the plants, and arranging the planting layout more rationally to avoid overcrowding or sparseness between plants, thereby optimizing land use efficiency and increasing yield per unit area.

[0108] The location information refers to the specific spatial coordinates or relative position description of the plant in the planting area. It may include the row or column position of the plant in the field, or the precise geographical location expressed in geographic coordinates (such as latitude and longitude), or the orientation and distance relative to a specific reference point (such as field ridges, irrigation facilities, etc.). Optionally, the method, tool or algorithm for analyzing the location information of the plant in the planting area can be used, such as: (and examples of the methods, tools or algorithms that can be used).

[0109] Based on the location information, this invention can invert the soil moisture content corresponding to the planting area, and can carry out targeted irrigation to avoid water waste and soil salinization caused by over-irrigation, and also prevent insufficient irrigation from affecting crop growth, thereby improving water resource utilization efficiency and crop yield.

[0110] The soil moisture content refers to the amount of water contained in a unit volume or unit mass of soil. Commonly used methods of expression include weight moisture content and volumetric moisture content, which are used to describe the degree of soil moisture.

[0111] As an embodiment of the present invention, the step of retrieving the soil moisture content corresponding to the planting area based on the location information includes: querying the geographic coordinates in the location information; analyzing the land use type corresponding to the planting area based on the geographic coordinates; determining the coverage area corresponding to the planting area based on the land use type; analyzing the soil moisture content ratio in the planting area based on the coverage area; and retrieving the soil moisture content corresponding to the planting area based on the soil moisture content ratio.

[0112] The geographical coordinates refer to a set of values ​​representing the location of a point in the planting area using latitude and longitude, consisting of two values: longitude and latitude. The land use type refers to the way and purpose in which the land is used, such as arable land, forest land, grassland, construction land, and water area, reflecting the main function and use of the land in the planting area. The coverage area refers to the geographical area occupied by the planting area, including its boundaries and size. The soil moisture content refers to the proportion of water content in the soil to the total volume or mass of the soil.

[0113] Furthermore, the querying of geographic coordinates in the location information can be achieved using GIS tools, such as ArcGIS and QGIS; the analysis of the land use type corresponding to the planting area can be achieved using remote sensing processing tools, such as ENVI and ERDAS; the determination of the coverage area corresponding to the planting area can be achieved using online map tools, such as the developer interfaces of Baidu Maps and Gaode Maps; the analysis of the soil moisture content ratio in the planting area can be achieved using TDR tools, such as using a TDR device to measure the dielectric constant of the soil on-site to estimate the moisture content ratio; and the inversion of the soil moisture content corresponding to the planting area can be achieved using machine learning frameworks, such as TensorFlow and Scikit-learn.

[0114] By identifying the distribution characteristics of soil moisture content, this invention can understand the distribution of soil moisture in different regions, enabling the development of targeted irrigation plans to ensure that crops in each region receive adequate water supply, avoiding over-irrigation or under-irrigation, thereby improving crop yield and quality.

[0115] The content distribution characteristics refer to the variation patterns and features of soil moisture content within the spatial range of the planting area. This includes the differences in soil moisture content at different locations, the uniformity of distribution, the existence of concentrated high or low moisture areas, the trend of moisture content variation with depth, and specific distribution patterns in different terrains, soil types, and crop planting areas. Optionally, the identification of the content distribution characteristics corresponding to the soil moisture content can be achieved by interpolation algorithms, such as Kriging interpolation algorithm, inverse distance weighted interpolation algorithm, etc.

[0116] Furthermore, based on the aforementioned content distribution characteristics, this invention generates an inversion report of soil moisture in the planting area, which can clearly understand the details of moisture distribution in each area of ​​the entire planting area, thereby accurately formulating irrigation plans, avoiding water waste and increased costs caused by blind watering, and ensuring that crops receive sufficient and appropriate water supply during critical growth periods, promoting good crop growth and high yield.

[0117] The inversion report refers to a comprehensive written or electronic document formed based on the analysis and calculation of soil moisture content distribution characteristics. It contains detailed information about the soil moisture status of the planting area, such as specific moisture content values ​​in different areas, description and analysis of distribution patterns, comparison with historical data or standard values, discussion of factors that may affect moisture distribution, and conclusions and suggestions based on these analyses. Optionally, the generation of the soil moisture inversion report of the planting area based on the content distribution characteristics can be achieved by report generation tools, such as Power BI, Tableau, etc.

[0118] First, by acquiring relevant field data corresponding to the planting area, this invention can comprehensively and accurately understand the actual conditions of the planting area, providing rich and reliable basic information for soil moisture inversion and providing a basis for subsequent analysis of the relationship between soil moisture and vegetation growth, thereby improving the accuracy of the inversion results. Simultaneously, by preprocessing the relevant field data, this invention obtains preprocessed field data, which can remove noise and outliers, improving data quality and accuracy. It can also unify and standardize data from different sources and formats, facilitating data integration and comprehensive utilization. Based on a preset microwave scattering mechanism model, this invention performs backscattering simulations on the growth stages of different plant species in the planting area, obtaining backscattering simulation parameters, which helps to deeply understand the electromagnetic scattering characteristics of crops at different growth stages. This invention can predict the scattering patterns of microwaves by plants at different growth stages in advance, providing a theoretical basis for subsequent remote sensing monitoring and data analysis. This allows for more accurate interpretation of information in remote sensing images and improves the precision of crop monitoring. By querying the time observation sequence in the actual observation values, this invention helps to explore the influence mechanism of environmental factors and plant growth stages on backscattering, revealing patterns related to plant growth cycles, meteorological cycles, or other periodic phenomena. This provides a strong basis for prediction and model optimization. Furthermore, based on the multi-source feature data, this invention analyzes the location information of plants in the planting area, enabling precise planning of the planting area. Understanding the location distribution of plants allows for a more rational arrangement of the planting layout, avoiding overcrowding or sparseness between plants, thereby optimizing land use efficiency and increasing yield per unit area. Therefore, the soil moisture inversion method and system based on microwave scattering mechanism and DNN model proposed in this invention can improve the effect of soil moisture inversion.

[0119] like Figure 2 The diagram shown is a functional block diagram of a soil moisture retrieval system based on microwave scattering mechanism and DNN model provided in an embodiment of the present invention.

[0120] The soil moisture retrieval system 200 based on microwave scattering mechanism and DNN model described in this invention can be installed in an electronic device. Depending on the functions implemented, the soil moisture retrieval system 200 may include a field data module 201, a crop parameter module 202, an observation value calculation module 203, a feature extraction module 204, and a report generation module 205. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0121] In this embodiment, the functions of each module / unit are as follows:

[0122] The field data module 201 is used to acquire the field-related data corresponding to the planting area. The field-related data includes: dual-polarization spaceborne SAR data, multispectral satellite remote sensing data, and field observation data.

[0123] The crop parameter module 202 is used to preprocess the field data to obtain preprocessed field data, extract the bipolar crop index from the preprocessed field data, query the surface crop parameters corresponding to the bipolar crop index, and determine the growth period of different plant species in the planting area based on the surface crop parameters.

[0124] The observation value calculation module 203 is used to perform backscattering simulation on the growth period of different plant species in the planting area based on a preset microwave scattering mechanism model, obtain backscattering simulation parameters, and calculate the actual observation value corresponding to the backscattering simulation parameters.

[0125] The feature extraction module 204 is used to query the time observation sequence in the actual observation value, combine the preset DNN model with the time observation sequence, construct the multi-source time series feature dataset corresponding to the planting area, and extract the multi-source feature data in the multi-source time series feature dataset.

[0126] The report generation module 205 is used to analyze the location information of the plants in the planting area based on the multi-source feature data, invert the soil moisture content corresponding to the planting area based on the location information, identify the content distribution characteristics corresponding to the soil moisture content, and generate an inversion report of soil moisture in the planting area based on the content distribution characteristics.

[0127] In detail, each module in the soil moisture inversion system 200 based on microwave scattering mechanism and DNN model described in this embodiment of the invention uses the same technical means as the soil moisture inversion method based on microwave scattering mechanism and DNN model described in the accompanying drawings, and can produce the same technical effect, which will not be repeated here.

[0128] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A soil moisture inversion method based on a microwave scattering mechanism and a DNN model, characterized in that, The method comprises: acquiring corresponding field-related data of the to-be-planted area, wherein the field-related data comprises dual-polarized spaceborne SAR data, multispectral satellite remote sensing data, and field observation data; performing data preprocessing on the field-related data to obtain preprocessed field data, extracting a dual-polarized crop index in the preprocessed field data, querying a corresponding ground crop parameter of the dual-polarized crop index, and determining a corresponding growth period of different plants in the to-be-planted area based on the ground crop parameter; based on a preset microwave scattering mechanism model, performing backscattering simulation on the growth period of different plants in the to-be-planted area to obtain backscattering simulation parameters, and calculating actual observation values corresponding to the backscattering simulation parameters; querying a time observation sequence in the actual observation values, combining a preset DNN model with the time observation sequence, constructing a multi-source time sequence feature data set corresponding to the to-be-planted area, and extracting multi-source feature data in the multi-source time sequence feature data set; based on the multi-source feature data, analyzing position information of plants in the to-be-planted area, based on the position information, inverting soil water content corresponding to the to-be-planted area, identifying content distribution characteristics corresponding to the soil water content, and generating an inversion report of soil water content in the to-be-planted area based on the content distribution characteristics; the extraction of the dual-polarized crop index in the preprocessed field data comprises: performing data unit segmentation on the preprocessed field data to obtain data segmentation units; extracting horizontal polarization data and vertical polarization data in the data segmentation units; performing filtering processing on the horizontal polarization data and the vertical polarization data to obtain horizontal polarization filtering data and vertical polarization filtering data; based on the horizontal polarization filtering data and the vertical polarization filtering data, calculating a dual-polarized ratio corresponding to the preprocessed field data; the dual-polarized ratio corresponding to the preprocessed field data is calculated by using the following formula: wherein, represents a data value corresponding to the horizontal polarization filter data, represents a weight coefficient corresponding to the horizontal polarization filter data, represents a weight coefficient corresponding to the horizontal polarization filter data, represents an offset corresponding to the horizontal polarization filter data, represents a constant term corresponding to the horizontal polarization filter data, represents a data value corresponding to the vertical polarization filter data, represents a weight coefficient corresponding to the vertical polarization filter data, represents an offset corresponding to the vertical polarization filter data, represents a constant term corresponding to the vertical polarization filter data. based on the dual-polarized ratio, determining the dual-polarized crop index in the preprocessed field data; the backscattering simulation based on the preset microwave scattering mechanism model on the growth period of different plants in the to-be-planted area to obtain backscattering simulation parameters comprises: collecting detailed parameters of the growth period of different plants in the to-be-planted area; based on the detailed parameters, performing parameter setting on the preset microwave scattering mechanism model; and using the parameter-set microwave scattering model, constructing a crop framework corresponding to the to-be-planted area; identifying crop framework parameters corresponding to the crop framework; performing backscattering simulation on the crop framework parameters to obtain backscattering simulation parameters.

2. The soil moisture inversion method based on microwave scattering mechanism and DNN model according to claim 1, wherein, the determination of the growth period of different plants in the to-be-planted area based on the ground crop parameter comprises: performing parameter screening on the ground crop parameter to obtain crop screening parameters; identifying crop characteristics corresponding to the crop screening parameters, querying crop characteristic indexes corresponding to the crop characteristics; based on the crop characteristic indexes, analyzing a period stage corresponding to the ground crop parameter; based on the period stage, determining the growth period of different plants in the to-be-planted area.

3. The soil moisture inversion method based on microwave scattering mechanism and DNN model according to claim 1, wherein, The query time observation sequence in the actual observation value includes: Collecting a complete data set corresponding to the actual observation value, and extracting an observation field in the complete data set; Performing standardization processing on the observation field to obtain a standardized field; Labeling a field index corresponding to the standardized field, and setting a time interval corresponding to the field index; Based on the time interval, query the time observation sequence in the actual observation value.

4. The soil moisture inversion method based on microwave scattering mechanism and DNN model according to claim 1, wherein, The preset DNN model is combined with the time observation sequence to construct a multi-source time sequence feature data set corresponding to the to-be-planted area, including: Identifying the adaptive format of the data in the preset DNN model; Based on the adaptive format, adjust the sequence observation data corresponding to the time observation sequence; Extracting multi-source sequence features from the sequence observation data; Perform vector conversion on the multi-source sequence features to obtain a multi-source feature vector; Using a preset DNN model to perform time sequence processing on the multi-source feature vector to obtain a multi-source time sequence feature data set corresponding to the to-be-planted area.

5. The soil moisture inversion method based on microwave scattering mechanism and DNN model according to claim 1, wherein, The soil moisture content corresponding to the to-be-planted area is inverted based on the position information, including: Querying the geographic coordinates in the position information; Based on the geographic coordinates, analyze the land use type corresponding to the to-be-planted area; Based on the land use type, determine the coverage range corresponding to the to-be-planted area; Based on the coverage range, analyze the soil water content ratio in the to-be-planted area; Based on the soil water content ratio, the soil moisture content corresponding to the to-be-planted area is inverted.

6. A soil moisture inversion system based on microwave scattering mechanism and DNN model, characterized in that, The system for performing the soil moisture inversion method based on the microwave scattering mechanism and the DNN model as claimed in any one of claims 1-5, the system comprising: A field data module for obtaining field-related data corresponding to a to-be-planted area, the field-related data including: dual-polarized satellite SAR data, multi-spectral satellite remote sensing data, and field observation data; A crop parameter module for performing data preprocessing on the field-related data to obtain preprocessed field data, extracting a dual-polarized crop index from the preprocessed field data, querying ground crop parameters corresponding to the dual-polarized crop index, and determining growth periods of different plants in the to-be-planted area based on the ground crop parameters; An observation value calculation module for performing backscattering simulation on the growth periods of different plants in the to-be-planted area based on a preset microwave scattering mechanism model to obtain backscattering simulation parameters and calculate actual observation values corresponding to the backscattering simulation parameters; A feature extraction module for querying a time observation sequence in the actual observation value, combining a preset DNN model with the time observation sequence to construct a multi-source time sequence feature data set corresponding to the to-be-planted area, and extracting multi-source feature data from the multi-source time sequence feature data set; A report generation module for analyzing position information of plants in the to-be-planted area based on the multi-source feature data, inverting soil moisture content corresponding to the to-be-planted area based on the position information, identifying content distribution characteristics corresponding to the soil moisture content, and generating an inversion report of soil moisture in the to-be-planted area based on the content distribution characteristics.

Images