A soil moisture standard quantitative detection method and a soil moisture control method

Abstract

The application provides a soil moisture standard quantitative detection method and a soil moisture control method, and belongs to the technical field of remote sensing, and comprises the following steps: defining a soil moisture standard quantitative product; the soil moisture standard quantitative product has unified specifications and quality requirements, so that different satellite soil moisture remote sensing products have consistency and comparability; a three-value evaluation system of the soil moisture standard quantitative product is constructed, and the three values include true values, agreed true values and relative true values; a soil moisture control method is constructed, the soil moisture control method is used as the relative true value of the soil moisture standard quantitative product; and the difference between the soil moisture standard quantitative product and the soil moisture control method is compared as a real test index. The application comprehensively unifies the real test method of the space-time frame parameter field, the standard quantitative product and the true value and the like concepts, so as to solve the consistency and equivalence problems of the existing soil moisture remote sensing products and improve the application level of remote sensing satellites.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of remote sensing, and particularly relates to the definition, construction method and authenticity testing technology of soil moisture standard quantitative product, especially aiming at the quality consistency and equivalence improvement of multi-source satellite remote sensing soil moisture product, a soil moisture standard quantitative detection method and a soil moisture control method are proposed. BACKGROUND

[0002] Soil moisture is an important part of the global water cycle system, and it is of great significance to agricultural production, ecosystem health and climate change research. Since the 1970s, a series of soil moisture product algorithms based on different satellite sensors (such as passive microwave, radar remote sensing, thermal infrared optical, etc.) have been developed. These products have different spatial resolutions from 1 meter to 100 meters or even global scale coarse grid resolution (such as 9 kilometers, 25 kilometers), providing users with a variety of choices. However, due to the influence of various error sources (such as algorithm, sensor and physical process), the quality of these satellite soil moisture products is quite different, each has its advantages and disadvantages. For example, there are many differences between different satellites in spatial resolution, spectral band setting, etc., making it difficult to carry out ground test data collection for each product, which brings challenges to the consistency and equivalence of soil moisture remote sensing products. The existing technology has the following problems:

[0003] High cost of data acquisition: Although the ground measured data has high accuracy, the cost of acquisition is high, and the site distribution is sparse and uneven, which is difficult to meet the demand of large-scale and continuous monitoring. The ground verification method is costly and difficult to match the satellite pixel scale; different satellites have different sensor parameters, inversion algorithms and error sources (such as surface heterogeneity), resulting in insufficient product consistency.

[0004] Complex multi-source data fusion: Traditional ground measurement methods (such as TDR, cosmic ray neutron instrument) are costly and difficult to match the satellite pixel scale; there are differences in resolution, time resolution and physical meaning between different data sources, and problems such as scale conversion and data assimilation need to be solved in the fusion process, which is technically difficult. The specifications of multi-source soil moisture products are not unified, and they cannot be directly fused and applied.

[0005] Limited model simulation accuracy: Model simulation requires a large amount of auxiliary data, and is strongly dependent on underlying conditions such as soil type and vegetation cover, so the accuracy and reliability of the simulation results are limited. There is a lack of standardized true value definition and testing index system; there is a lack of unified standardized testing framework, which limits the interoperability and business application of the product. SUMMARY

[0006] The application aims to provide a soil moisture standard quantitative detection method and a soil moisture control method, which comprehensively unifies the authenticity verification method of the space-time frame parameter field, the standard quantitative product and the true value, to solve the consistency and equivalence of the existing soil moisture remote sensing product and improve the application level of remote sensing satellites.

[0007] A soil moisture standard quantitative detection method comprises the following steps:

[0008] Step 1: defining a soil moisture standard quantitative product, the soil moisture standard quantitative product has unified specifications and quality requirements, so that different satellite soil moisture remote sensing products have consistency and comparability;

[0009] Step 2: constructing a three-value evaluation system of the soil moisture standard quantitative product, the three values include true value, agreed true value and relative true value, wherein the true value is the theoretical value, the agreed true value under certain measurement conditions is the relative true value, and the relative true value is used to replace the true value;

[0010] Step 3: constructing a soil moisture control method, taking the soil moisture control method as the relative true value of the soil moisture standard quantitative product;

[0011] Step 4: taking the difference between the soil moisture standard quantitative product and the soil moisture control method as the authenticity verification index, which is used to quantitatively evaluate the accuracy and uncertainty of the soil moisture standard quantitative product.

[0012] A soil moisture standard quantitative detection device comprises:

[0013] A definition module defines a soil moisture standard quantitative product, the soil moisture standard quantitative product has unified specifications and quality requirements, so that different satellite soil moisture remote sensing products have consistency and comparability;

[0014] An evaluation module constructs a three-value evaluation system of the soil moisture standard quantitative product, the three values include true value, agreed true value and relative true value, wherein the true value is the theoretical value, the agreed true value under certain measurement conditions is the relative true value, and the relative true value is used to replace the true value;

[0015] A soil moisture control method construction module constructs a soil moisture control method, taking the soil moisture control method as the relative true value of the soil moisture standard quantitative product;

[0016] An authenticity verification module takes the difference between the soil moisture standard quantitative product and the soil moisture control method as the authenticity verification index, which is used to quantitatively evaluate the accuracy and uncertainty of the soil moisture standard quantitative product.

[0017] An electronic device, comprising: one or more processors; a memory storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method described above.

[0018] A computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the method described above.

[0019] A soil moisture control block construction method, comprising:

[0020] Explicitly specify the specification requirements of the soil moisture standard quantitative product;

[0021] According to different specification requirements, the soil moisture control block is divided into different levels;

[0022] Select the corresponding area for sampling to design the sample plot;

[0023] According to the designed sample plot, measure and obtain the soil moisture data;

[0024] According to the characteristics and specification requirements of the soil moisture data, select a suitable model to construct different levels of soil moisture control blocks to constitute the sample plot;

[0025] By comparing with the known true value data, evaluate the soil moisture control block, and optimize and adjust the model to improve the quality of the soil moisture control block.

[0026] A soil moisture control block construction device, comprising:

[0027] The specification requirement specifying module specifies the specification requirements of the soil moisture standard quantitative product;

[0028] The level division module divides the soil moisture control block into different levels according to different specification requirements;

[0029] The sample plot design module selects the corresponding area for sampling to design the sample plot;

[0030] The data measurement and acquisition module measures and obtains the soil moisture data according to the designed sample plot;

[0031] The soil moisture control block construction module selects a suitable model to construct different levels of soil moisture control blocks to constitute the sample plot according to the characteristics and specification requirements of the data;

[0032] The optimization and adjustment module compares with the known true value data, and optimizes and adjusts the model to improve the quality of the soil moisture control block.

[0033] An electronic device comprising: one or more processors; memory storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to carry out the method described above.

[0034] A computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to carry out the method described above.

[0035] The present application has the following beneficial technical effects:

[0036] (1) The soil moisture standard quantitative product definition proposed in the present application unifies the specifications and quality requirements of the product, and solves the problem of large differences in quality of different satellite soil moisture products, which makes it difficult to compare consistency and equivalence. Through the standardized product definition, the soil moisture data obtained by different satellites can be processed and analyzed under the same framework, and a soil moisture standard quantitative product definition method with unified specification requirements and quality requirements is proposed to ensure the consistency and comparability of different satellite soil moisture products. Through multi-source data fusion and machine learning model verification, the accuracy is improved, the consistency and equivalence of remote sensing products are solved, the comparability and interoperability of data are improved, and a solid data foundation is provided for large-scale, multi-field remote sensing applications.

[0037] (2) The constructed soil moisture control party serves as a relative true value. The soil moisture control party with clear levels is constructed as the relative true value for the verification of the soil moisture standard quantitative product, which improves the accuracy and reliability of the verification data. It provides an objective and accurate benchmark for the verification of the soil moisture standard quantitative product. Compared with the true value uncertainty in the prior art which is limited by measurement means and methods, the control party construction method of the present application is more systematic and comprehensive, supports multi-level resolution from 1m to 1000m, adapts to different satellite data, fully considers various factors such as soil type, terrain, and vegetation coverage, and ensures the representativeness and accuracy of the control party. Through comparative analysis of the control party and remote sensing products, the error and bias of the product can be more effectively evaluated, which provides a clear direction for the improvement and optimization of the product, thereby improving the quality and application level of remote sensing products.

[0038] (3) The application provides a variety of methods for constructing soil moisture control fields, including based on ground measured data, multi-source data fusion model and intelligent model, etc. These methods have their own advantages and can be flexibly selected according to different product specifications and application scene requirements. Compared with the prior art, this method system is more perfect, can adapt to diversified data sources and complex ground conditions, and improves the feasibility and efficiency of control field construction. At the same time, when using intelligent model to construct control field, the advantages of machine learning algorithm can be fully played, the most relevant features can be automatically selected, the nonlinear relationship between remote sensing data and soil moisture can be captured, efficient soil moisture estimation can be carried out, and the accuracy and stability of the control field are further improved. For example, using machine learning to reduce the ground sampling density (such as 1000m resolution only needs 1 control field).

[0039] (4) The application defines the definitions of true value, agreed true value and relative true value of soil moisture standard quantitative product, and gives specific authenticity verification index calculation method. These indexes can comprehensively and quantitatively evaluate the accuracy and uncertainty of the product, and provide a scientific basis for product quality control and user selection. Compared with the prior art, the verification index of the application is more systematic and comprehensive, can more accurately reflect the error size and correlation with true value of the product, and is helpful to improve the credibility of remote sensing product and in-depth application in various fields. Providing authenticity verification index, defining product true value related definition, giving product verification index, providing scientific basis for quality evaluation of soil moisture remote sensing product. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 The figure is a schematic diagram of the data block of the application;

[0041] Figure 2 The figure is a flow chart of constructing soil moisture control field of the application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical scheme and advantages of the application more clear and understandable, the application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and not to limit the application. In addition, the technical features involved in each embodiment of the application described below can be combined with each other as long as they do not conflict with each other. In order to achieve the above purpose, the application adopts the following technical scheme.

[0043] The application provides a soil moisture standard quantitative detection method, which comprises the following steps:

[0044] Step 1: definition of soil moisture standard quantitative product; including:

[0045] Soil moisture standard quantitative products are quantitative data sets of soil moisture defined based on remote sensing technology, unified specifications and quality requirements. Using observation data obtained from different satellite sensors (such as passive microwave, radar remote sensing, thermal infrared optical, etc.), these data are converted into quantitative information reflecting soil moisture content through remote sensing inversion algorithms. These products usually describe the spatial variation details of soil moisture at a specific spatial resolution (such as 1m to 1000m), meeting the fine needs of soil moisture information for different application scenarios. Its application scenarios are widely used, including irrigation decision support in agricultural production, water cycle analysis in weather forecasting, and vegetation growth condition assessment in ecological environment monitoring. In terms of computer hardware and internal performance, its processing process has certain requirements for computing resources, especially in the processing of large amounts of remote sensing data and complex algorithm operations, high-performance computing equipment is needed to ensure processing efficiency and accuracy.

[0046] As an important part of the land surface physicochemical parameter product system, the production process of soil moisture standard quantitative products (SSMP) follows the transformation path of "radiometric calibration→ parameter inversion→ spatio-temporal reconstruction". Based on optical microwave radiation product input, it generates soil volume moisture data cubes with clear physical meaning through inversion algorithms driven by physical mechanisms. SSMP adopts a standardized process design, effectively eliminates algorithm exogenous differences through error propagation control models in production links, and ensures that the spatial and temporal consistency of product specifications meets the engineering constraint conditions of Δx≤0.05° and Δt≤6h.

[0047] (1) Variety setting

[0048] Soil moisture standard quantitative products are based on the interaction mechanism of electromagnetic waves and soil medium, and four types of standardized product lineages are constructed: (1) Optical remote sensing products based on spectral absorption characteristics in the visible-near infrared band; (2) Passive microwave remote sensing products relying on the penetration characteristics of radiometers to realize soil moisture inversion; (3) Active microwave remote sensing products using the response relationship between the backscattering coefficient and the dielectric constant of C / X-band synthetic aperture radar; (4) Multi-source fusion products integrating multiple physical mechanism inversion results through the D-S evidence theory framework.

[0049] (2) Specification setting

[0050] The spatiotemporal reference system adopts a double-constrained standardized framework: (1) spatial standard field, based on a five-layer fifteen-level spatial resolution hierarchy model, through a double-latitude-longitude grid system to achieve 0.01°-1° multi-scale spatial reference; (2) time standard field, UTC time coding system is constructed, and sliding time window algorithm is adopted to realize seamless connection of daily, ten-day and inter-annual scale products. The specification system realizes seamless integration of multi-source data through geographic space conversion library.

[0051] (3) Quality requirements

[0052] A four-element quality evaluation model Q = (A, P, U, C) is established, where: accuracy (Accuracy) contains the absolute accuracy requirement of RMSE≤0.04 m 3 / m 3 ; reliability (Precision) is measured by the correlation coefficient R 2 ≥0.85 between cross-validation products; uncertainty (Uncertainty) is quantified by the Monte Carlo method to propagate the error of the inversion process; consistency (Consistency) requires Bias≤±0.02 m 3 / m 3 in the overlapping area of multi-source products. The quality evaluation process establishes the value traceability chain from the sensor to the terminal product.

[0053] (4) Timeliness and scale requirements

[0054] The timeliness index adopts a time response function, and requires that the business product meet the lag threshold of τ≤12 h (daily product) and τ≤72 h (ten-day product). It effectively supports the cross-scale application demand from local hydrological model assimilation to global water cycle research.

[0055] Step 2: Definition and construction of true value, agreed true value and relative true value of soil moisture standard quantitative products; including:

[0056] The true value of the standard quantitative product of soil moisture refers to the true connotation and appearance of soil moisture at the pixel scale, which is the objective and real-world data, and does not depend on the data and algorithm of remote sensing products. Due to the difficulty of obtaining absolute true value in actual measurement, the concepts of conventional true value and relative true value are introduced. The conventional true value is defined as the standard true value, which is constantly approaching the true value through measurement method improvement, has a clear sampling specification, and its value changes with equipment and observation process. The level of technology can change the value of the conventional true value. The conventional true value is obtained within the ideal number of observations with a certain allowable deviation. This true value is reachable and infinitely close to the true value, but it cannot reach the true value definition, and it has a certain degree of uncertainty. It can be obtained by multiple sampling using observation methods without systematic errors. The relative true value is based on the existing measurement means and tries to accurately approach the absolute true value. The methods to construct these true values include ground-based measurement data, multi-source data fusion model, and intelligent model. For example, based on ground-based measurement data, the ground multi-point observation value or footprint / patch observation value is directly used, but this method is limited by factors such as observation cost and site distribution. The multi-source data fusion model integrates different precision and quality data, such as other satellite soil moisture products, terrain, vegetation products, and ground observations, to construct more accurate control fields. The intelligent model uses machine learning algorithms to model the nonlinear relationship between multi-source remote sensing data and ground observation data, generating spatially continuous soil moisture control true values.

[0057] (1) True value

[0058] The observed object value obtained by remote sensing is the value obtained by sampling at a certain spatial and temporal scale. The parameter field value with a clear spatio-temporal range consistent with the standard quantitative product specification can be defined as the true value of the standard quantitative product. This true value is the intrinsic property value of the measured object in the ideal undisturbed state, which meets the absolute objectivity, mathematical uniqueness, and physical determinacy, and is the convergence limit of all actual measurement values.

[0059] The true value is the true connotation and appearance of soil moisture at the pixel scale, which is objectively existing and real-world data, and does not depend on any remote sensing product data and algorithm. The true value not only reflects the spatio-temporal field characteristics of soil moisture, but also reflects the product specification requirements, with multiple characteristics such as definition, authenticity, accuracy, and feasibility. Therefore, the true value with soil moisture spatio-temporal sampling characteristics and consistent with product specification requirements is the true value of the standard quantitative product of soil moisture.

[0060] (2) Conventional true value

[0061] In practical work, through the agreement standard specification, the soil moisture spatiotemporal information set obtained by different measurement methods can be optimized, or a certain certified measurement method is specified, for example, the data accuracy of the pixel true value is obtained multiple times within a certain error range, so as to determine the true value of the soil moisture product at the pixel scale, that is, the agreed true value, which is the measurable and accurate soil moisture spatiotemporal distribution characteristic value. We regard this agreed true value as the true value. The agreed true value is a true value at the experimental level, which has accuracy and accessibility, and is used as a substitute for true value in practical work.

[0062] The agreed true value is defined as the standard true value, and is constantly approaching the true value through the improvement of measurement methods. Its value changes with equipment and observation process, and the improvement of technical level can change the value of the agreed true value. The agreed true value is obtained within the allowed deviation and ideal observation times. Such agreed true value has accessibility and approaches the true value, but cannot reach the theoretical true value, and has uncertainty to a certain extent, that is, theoretically, the observation method without systematic error is used to obtain through multiple sampling. Its simplest form can be expressed by a mathematical equation:

[0063] (1)

[0064] (2)

[0065] The true value of the soil moisture standard quantitative product is represented by The observation data obtained by a certain observation means is represented by β, which represents the weight factor, and t represents the infinite sampling under ideal conditions.

[0066] (3) Relative true value

[0067] Ground measurement data is the main source of pixel scale true value in true value test, but limited by measurement means and measurement methods, absolute true value cannot be obtained, only the best approximation value. The relative true value of remote sensing standard quantitative product is to obtain the true value as close as possible to the absolute true value. Like the agreed true value, such true value has accessibility and approaches the true value, but cannot reach the true value definition, and has uncertainty to a certain extent.

[0068] In actual measurement, different true values may appear due to different measurement conditions, such as different measurement methods, measurement equipment, measurement personnel, and measurement environment. This also illustrates the unmeasurability of the true value definition, and only a conventional true value under certain measurement conditions exists, that is, the relative true value commonly used at present. In the limit case, the average value of ideal measurement values without system error and sufficient times can be considered as the pixel-level conventional true value of the standard quantitative product. In limited actual measurement, both systematic errors and accidental errors exist, and there is always a difference between the measured value and the actual value, which is the error. The relative true value needs to be improved by continuously improving the measurement method to improve its accuracy, and its value will change with the improvement of the method.

[0069] In summary, the conventional true value has reachability, and the relative true value has measurability. The relative true value is an optimal solution under actual conditions, which is equivalent to a higher precision observation, and is generally used as the true value for analysis and evaluation of remote sensing information products. The relative true value of soil moisture as a reference is a kind of measurement true value under semi-restrained intensity between indoor physical experiment and field experiment.

[0070] Step 3: Construction of soil moisture control square; including:

[0071] Definition of "five-control" square; including: Data Square (abbreviated as DS) is a geographical spatio-temporal grid data organization model based on the grid division of the earth's surface and the mapping of specific index data to grid cells. Its core is to build a unified geographical grid framework, and through the spatio-temporal relationship (time sequence, spatial distribution), attribute semantic association and evolution process, etc. Multi-dimensional information network, standardized integration of multi-source spatio-temporal data. This structured organization mode significantly improves the efficiency of computer artificial intelligence processing and big data analysis, so it is called "geographical spatio-temporal grid data wisdom cube". Data Square (Data Square) is shown in Figure 1 The figure shows how to divide the earth's surface according to the grid, label different resolution levels (such as 1m, 100m, 1000m), and form index data mapping, reflecting the size of the grid and the data organization structure under different resolutions.

[0072] Remote sensing sampling can generate multi-level grid squares from 10 -4 degrees to 10 degrees (about 1 centimeter to 1000 meters on the equator) by discretizing the continuous surface through limited resolution imaging. Its spatial resolution adopts a standardized hierarchical system: 0.1m, 0.25m, 0.5m, 1m, 2.5m, 5m, 10m, 25m, 50m, 100m, 250m, 500m, 1000m, 2500m, 5000m.

[0073] The definition of soil moisture control square is:

[0074] Based on the remote sensing inversion technology of multi-source satellite data such as optical / near-infrared, thermal infrared and active / passive microwave, the dynamic monitoring of soil moisture is realized. However, the spatial resolution of multi-satellite load is significantly different, which leads to the difficulty of matching each product with ground test. Therefore, a standardized and multi-level soil moisture ground reference verification sample system, i.e. soil moisture control field, needs to be built.

[0075] Soil moisture control field is a high-precision relative true value generated by ground measurement and data processing. Based on known geographic coordinates and geometric shape, it represents the soil moisture field with spatial distribution characteristics, and is used to quantitatively evaluate the deviation of remote sensing products and optimize the inversion algorithm through authenticity test to approach the true value. The resolution level of control field is set to 1m, 5m, 10m, 25m, 50m, 100m, 250m, 500m, 1000m, etc. according to geographic grid and main satellite parameter, which has the characteristics of standardization and full coverage.

[0076] With the progress of intelligent observation technology, soil moisture measurement equipment has been expanded from point scale (such as TDR) to area scale (cosmic ray observation instrument, hundreds of meters) and kilometer scale (LAS). When the sample area is divided into m control field units, the true average moisture content can be represented by the average value of each unit. For example: when verifying the SSMP product with 1000m spatial resolution, theoretically, 106 1m control fields (1000x1000), 104 10m control fields (100x100), 102 100m control fields (10x10) or 1 1000m control field are needed. The scale of control field is inversely proportional to the sampling frequency, that is, the larger the scale, the fewer the sampling units needed.

[0077] Soil moisture control field is an objective entity representing the true state of pixel-scale soil moisture, and its attributes are independent of remote sensing product data and inversion algorithm. The following principles should be followed when building control field: soil moisture control field should fully consider soil type, terrain, vegetation cover type and other factors, and select areas with representative underlying surface. It should cover the main land cover types in the local area (except water, ice and snow, urban and construction land), and the spatial range of control field should not be less than 3x3 times the spatial resolution of the SSMP to be tested. It is better to choose flat areas, and should avoid military restricted areas, airport air control areas and other dangerous areas, and try to avoid roads, shelter belts, high-voltage lines and other ground objects, while ensuring relatively convenient traffic conditions. The soil moisture control field is built with the observation value closest to the time when the SSMP satellite data passes through, which is recommended to be within ±24h. It is better to stagger after precipitation, after irrigation and soil freezing period. If conditions permit, auxiliary information should be obtained at the same time, including atmospheric temperature, humidity, wind speed and other meteorological elements within a short distance above the soil surface.

[0078] Soil moisture control field construction includes:

[0079] The following three types of core methods are recommended for constructing soil moisture control methods. The first type is the pixel-scale measured data construction method, which uses high-precision instruments (such as COSMOS cosmic ray observation instruments) to directly obtain 500m level control data, or uses multi-point observation values to construct control methods through point-to-area scale conversion; the second type is the multi-source data fusion modeling method, which is based on fine grid observation data and uses optimization algorithms to construct control methods; the third type is based on intelligent learning model construction.

[0080] (1) Construction based on pixel-scale ground measured data

[0081] Directly use ground multi-point observation values or footprint / patch observation values to construct soil moisture control methods. With the increase of pixel scale, the use of dense observation network measured data can obtain high reliability verification results, but this method has high observation cost, and in practice, ground observation stations are sparse and unevenly distributed. Moreover, in actual field sampling, under certain error and confidence conditions, uniform

underlying surface quadrat

[0082] The core of the true value test is the acquisition of true value data, and related methods continue to iterate and innovate: for example, including ground layout optimization based on spatial simulation annealing algorithm, geostatistical layout method, spatial non-uniform surface mean estimation sampling, hierarchical optimization sampling, and cross-validation reference dataset construction, etc.

[0083] This construction method is limited by measurement means and methods, and in actual true value test, the true value cannot be obtained, only the best approximation value can be obtained. The establishment of a typical ecological system soil automatic observation network (referred to as "space-based soil moisture network") covering major ecological systems (including grassland, farmland, desert and forest) and dry and wet climate zones, with unified data specifications and quality requirements, can obtain 100m scale soil moisture relative true value, which is a typical 100m level soil moisture control method.

[0084] (2) Construction based on multi-source data fusion model

[0085] With the increase of pixel scale, the heterogeneity of the ground surface also increases. When the amount of ground measured data is small, it is difficult to represent the true value of the pixel scale. It is recommended to use non-homogeneous surface spatial inference model, regression kriging, Bayesian maximum entropy and other methods, and introduce auxiliary information with strong correlation with the estimated variable or similar spatial distribution pattern to improve the estimation accuracy. Under the support of relevant auxiliary information and high-resolution remote sensing observation, a hierarchical Bayesian model is constructed or a spatio-temporal kriging method is used to integrate various prior information and spatio-temporal observation data, and to realize accurate estimation of soil control and uncertainty evaluation under the constraint of variable time-varying law. This method can be used for large-scale spatial heterogeneity research, and can integrate various data of different precision and quality to construct 250m, 500m, and 1000m level soil control.

[0086] (3) Based on intelligent learning model construction

[0087] Remote sensing images usually have high dimension and large scale characteristics, and traditional statistical methods are difficult to process such complex data sets. Machine learning models (such as random forest, gradient boosting tree, etc.) can effectively process big data, quickly extract useful information from multi-dimensional data, and generate spatially continuous soil moisture control true value. The spatial distribution and dynamic change of soil moisture are affected by many factors, such as vegetation, terrain, climate, etc. Machine learning models can model the nonlinear relationship between multi-source remote sensing data and ground observation data, automatically select the most relevant features, and help researchers understand the main driving factors of soil moisture, and further optimize the model. At the same time, during the model training process, the precision and stability of the final product can be improved by continuously adjusting the hyperparameters and model structure.

[0088] When using machine learning models to construct soil moisture remote sensing product true value, first, multi-source remote sensing data and ground measured data are collected as model input and training target. At the same time, auxiliary data such as meteorological and topographic data are expanded. Model selection usually includes classic machine learning algorithms such as random forest and gradient boosting tree, which can improve model performance through feature selection and hyperparameter optimization. These models can capture the nonlinear relationship between remote sensing data and soil moisture, and perform efficient soil moisture estimation. Finally, the model accuracy is evaluated by cross-validation, and the spatially continuous soil moisture true value product is generated by interpolation method.

[0089] The soil moisture control construction process mainly consists of six stages, and the construction process is as follows Figure 2The six stages from requirement analysis to verification optimization are shown, and the specific steps and mutual relations are shown, and the arrow direction indicates the data flow direction, which helps to understand the logic and order of the whole construction process. The soil moisture control party construction flow chart contains the demand analysis, hierarchical standardization, model training and verification optimization module. First of all, the requirement analysis, the specification requirements of the soil moisture standard quantitative product are clear, such as spatial resolution, time resolution, product accuracy, etc. Secondly, the hierarchical standardization, according to the different resolution requirements, the control party is divided into different levels, such as 1m, 2.5m, 5m, etc. The sample plot design stage needs to consider the soil type, terrain, vegetation cover type and other factors, and select the representative area for sampling. In the data collection stage, according to the designed sample plot, various measurement means are used to obtain soil moisture data. In the control party model construction stage, according to the data characteristics and product requirements, the appropriate model is selected, such as the mean model based on ground measured data, multi-source data fusion model or intelligent model, to construct the soil moisture control party. Finally, the verification and optimization stage, through the comparison with the known true value data, the accuracy and reliability of the control party are evaluated, and the model is optimized and adjusted to improve the quality of the control party.

[0090] Step 4: SSMP authenticity test based on soil moisture control party; including:

[0091] Considering the specification requirements of the soil moisture standard quantitative product, the soil moisture control party constructed is used as the relative true value to carry out the satellite-ground verification of SSMP. The main operation process includes: carrying out synchronous measurement for the SSMP to be verified; determining the control party construction model according to the hierarchical scale and measurement data; making soil moisture control party of selected level; completing the test and evaluation of SSMP.

[0092] SSMP authenticity test indicators include accuracy evaluation indicators and uncertainty evaluation indicators;

[0093] Quality test includes numerical accuracy and reliability. Only when a certain degree of accuracy is reached, the information can meet its specification requirements, such as accurately reflecting the state and change law of the observed object. Moreover, this compliance must withstand the test of the entire transmission process, overcome the complexity, instability of the objective itself and the imperfection of information, system error and random error generated during data collection, processing and application analysis, to ensure the reliability in a certain space and time, with stable production process and traceability.

[0094] (1) Accuracy evaluation index

[0095] The available indicators for quantitatively expressing the accuracy of SSMP include: mean error (ME), mean absolute error (MAE), mean relative error (MRE), root mean square error (RMSE) and correlation coefficient (R). The calculation is shown in formulas (3)-(7).

[0096] (3)

[0097] (4)

[0098] (5)

[0099] (6)

[0100] (7)

[0101] (2) Uncertainty evaluation index

[0102] Using standard deviation ( The uncertainty of SSMP is quantitatively expressed by indicators such as covariance (COV). The calculation formulas for these indicators are shown in (8)-(9).

[0103] (8)

[0104] (9)

[0105] In the formula:

[0106] m i —The value of the i-th SSMP cell, in cm 3 cm -3 ,

[0107] d i —The i-th soil moisture control value, in cm 3 cm -3 ,

[0108] n — Number of samples, in units

[0109] —Mean of SSMP sample points, cm 3 cm -3 ,

[0110] —Mean value of soil moisture control data, in cm 3 cm -3 .

[0111] Authenticity testing indicators are used to quantitatively evaluate the accuracy and uncertainty of soil moisture standard quantitative products. Accuracy evaluation indicators include mean error (ME), mean absolute error (MAE), mean relative error (MRE), root mean square error (RMSE), and correlation coefficient (R). These indicators are calculated by comparing the differences between the pixel values ​​of the soil moisture standard quantitative product and the soil moisture control values, comprehensively reflecting the magnitude of the product's error and its correlation with the true value. Uncertainty evaluation indicators utilize the standard deviation (...). ), covariance (COV), etc., quantitatively express the uncertainty of the soil moisture standard quantitative product, and reflect the dispersion degree and stability of the data. In actual application, these indexes can help users understand the quality status of the product, provide a basis for improvement and optimization of the product, and also provide a reference for users to select appropriate products in different application scenarios.

[0112] The application further provides a soil moisture standard quantitative detection device, comprising:

[0113] The definition module defines the soil moisture standard quantitative product, and the soil moisture standard quantitative product has unified specifications and quality requirements, so that different satellite soil moisture remote sensing products have consistency and comparability.

[0114] The evaluation module constructs a three-value evaluation system of the soil moisture standard quantitative product, and the three values include true value, agreed true value and relative true value, wherein the true value is a theoretical value, the agreed true value under certain measurement conditions is the relative true value, and the relative true value is used to replace the true value.

[0115] The soil moisture control party construction module constructs a soil moisture control party, and uses the soil moisture control party as the relative true value of the soil moisture standard quantitative product.

[0116] The authenticity verification module uses the difference between the soil moisture standard quantitative product and the soil moisture control party as an authenticity verification index, and is used for quantitatively evaluating the accuracy and uncertainty of the soil moisture standard quantitative product.

[0117] The application further provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement a soil moisture standard quantitative detection method of the application.

[0118] The application further provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, cause the processor to implement a soil moisture standard quantitative detection method of the application.

[0119] On the other hand, the application further provides a soil moisture control party construction device, comprising:

[0120] The specification requirement clear module clearly defines the specification requirements of the soil moisture standard quantitative product.

[0121] The hierarchical division module divides the soil moisture control party into different levels according to different specification requirements.

[0122] The sample plot design module selects a corresponding area for sampling to design a sample plot.

[0123] A data measurement acquisition module acquires soil moisture data according to a designed sample plot;

[0124] A soil moisture control rule construction module selects a suitable model according to data characteristics and specification requirements to construct different levels of soil moisture control rules to form a sample plot.

[0125] An optimization adjustment module compares with known true value data, optimizes and adjusts the model, and improves the quality of the soil moisture control rules.

[0126] The application further provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the soil moisture control rule construction method.

[0127] The application further provides a computer readable storage medium, which stores executable instructions, and the instructions make the processor implement the soil moisture control rule construction method when executed by the processor.

Claims

1. A method for standard quantitative detection of soil moisture, characterized in that, Comprise the following steps: Step 1: define the soil moisture standard quantitative product, the soil moisture standard quantitative product has unified specifications and quality requirements, so that different satellite soil moisture remote sensing products have consistency and comparability; Step 2: build a three-value evaluation system for soil moisture standard quantitative products, the three values include true value, agreed true value and relative true value, wherein the true value is the theoretical value, the agreed true value under certain measurement conditions is the relative true value, and the relative true value is used instead of the true value; Step 3: build a soil moisture control method, and use the soil moisture control method as the relative true value of the soil moisture standard quantitative product; Step 4: the difference between the soil moisture standard quantitative product and the soil moisture control method is used as the authenticity test index for quantitative evaluation of the accuracy and uncertainty of the soil moisture standard quantitative product; The varieties of soil moisture standard quantitative products include: Optical remote sensing products based on the spectral absorption characteristics of visible and near-infrared bands; Passive microwave remote sensing products relying on the penetration characteristics of radiometers to realize soil moisture inversion; Active microwave remote sensing products using the response relationship between the backscattering coefficient and the dielectric constant of C / X band synthetic aperture radar; Multi-source fusion products integrating multi-physical mechanism inversion results; The quality evaluation method of soil moisture standard quantitative product is: Establish a four-element quality evaluation model Q=(A, P, U, C), wherein: A is the accuracy; P is the reliability; U is the uncertainty; C is the consistency; In step 2, the agreed true value is expressed by a mathematical equation: (1) (2) denotes the true value of the soil moisture standard quantitative product, denotes the observation data obtained by a certain observation means, β represents a weight factor, t represents infinite sampling under ideal conditions, and n represents the sample number.

2. The method according to claim 1, wherein, In step 1: the soil moisture standard quantitative product is a soil moisture quantitative data set defined based on remote sensing data according to unified specifications and quality requirements; the remote sensing data is data obtained by satellite sensors, which is converted into quantitative information reflecting soil moisture content by remote sensing inversion algorithm.

3. The method according to claim 1, wherein the soil water content is quantitatively determined by the method. The specifications of the soil moisture standard quantitative product include: Spatial standard field, based on spatial resolution level model, through double latitude and longitude grid to realize 0.01-1 multi-scale spatial reference; Time standard field, build UTC time coding system, use sliding time window algorithm to realize seamless connection of daily, ten-day and interannual scale products.

4. The method according to claim 1, wherein, The timeliness and scale requirements of soil moisture standard quantitative product are as follows: The timeliness index adopts time response function τ, the lag threshold of daily product τ≤12 h, ten-day product τ≤72 h.

5. The method according to claim 1, wherein the soil water content is quantitatively determined by the method. The soil moisture control method is a relative true value generated by ground measurement and data processing, which is based on known geographic coordinates and geometric shape, represents soil moisture field with spatial distribution characteristics, is used for quantitative evaluation of product deviation, and is used for real-time test to optimize inversion algorithm to approach true value.

6. The method of claim 1, wherein the soil water standard is selected from the group consisting of: The resolution level of soil moisture control method is set to 1m, 5m, 10m, 25m, 50m, 100m, 250m, 500m, 1000m and other levels according to geographic latitude and longitude grid and mainstream satellite parameters. ​ 7. The method of claim 1, wherein the soil water standard is selected from the group consisting of: Step 4 includes: the authenticity test index of soil moisture standard quantitative product includes accuracy evaluation index and uncertainty evaluation index; ​ The accuracy evaluation indexes include: mean error ME, mean absolute error MAE, mean relative error MRE, root mean square error RMSE and correlation coefficient R; the calculation is shown in formulas (3)-(7): (3) (4) (5) (6) (7) The uncertainty evaluation indexes include: standard deviation σ, covariance COV, and the index calculation formulas are shown in (8)-(9): (8) (9) In the formula: m i - the i-th soil moisture standard quantitative product pixel value, d i - the i-th soil moisture control function value, - the mean of the soil moisture standard quantitative product sample points, - Mean of soil moisture control side data.

8. The method according to claim 1, wherein, Wherein, The soil moisture control block is constructed by the following steps: Defining the specification requirements of the soil moisture standard quantitative product; According to different specification requirements, the soil moisture control block is divided into different levels; Selecting the corresponding area for sampling to design the sample block; According to the designed sample block, the soil moisture data is measured and obtained; The soil moisture control block is constructed, and according to the characteristics and specification requirements of the soil moisture data, a suitable model is selected to construct the soil moisture control block of different levels to form the sample block; The soil moisture control block is evaluated, and the model is optimized and adjusted to improve the quality of the soil moisture control block.

9. The method according to claim 8, wherein the soil water standard is a soil water content standard. The specification requirements include spatial resolution, temporal resolution and product precision.

10. The method of claim 8, wherein the soil water standard is selected from the group consisting of: The suitable model includes the mean model based on ground measured data, multi-source data fusion model or intelligent learning model. ​ 11. The method according to claim 10, wherein the soil water standard is a soil water content standard. The mean model based on ground measured data is used to construct the soil moisture control block; including: Directly using the multi-point observation value or footprint / patch observation value to construct the soil moisture control block; which is used to construct the 100m level soil moisture control block.

12. The method of claim 10, wherein the soil water standard is selected from the group consisting of: The multi-source data fusion model is used to construct the soil moisture control block; including: ​ Using the heterogeneous surface spatial inference model, regression kriging or Bayesian maximum entropy method, and introducing auxiliary information, constructing the hierarchical Bayesian model or using the spatio-temporal kriging method, comprehensively constructing the 250m, 500m and 1000m level soil control block under the constraint of variable time-varying law, and the auxiliary information includes the near distance atmospheric temperature, humidity and wind speed above the soil surface.

13. The method of claim 10, wherein the soil water standard is selected from the group consisting of: The intelligent learning model is used to construct the soil moisture control block; including: ​ Collecting multi-source remote sensing data and ground measured data as the input and training target of the intelligent learning model; at the same time, combining meteorological, terrain and other auxiliary data for expansion, selecting machine learning algorithm, improving the model performance through feature selection and hyperparameter optimization, finally evaluating the model precision by cross-validation, and generating spatially continuous soil moisture control block by interpolation method.

14. A soil moisture standard quantitative detection device, which implements the method of claim 1, characterized in that, Including: The definition module defines the soil moisture standard quantitative product, and the soil moisture standard quantitative product has unified specification and quality requirements, so that different satellite soil moisture remote sensing products have consistency and comparability; The evaluation module constructs a three-value evaluation system for the soil moisture standard quantitative product, and the three values include true value, agreed true value and relative true value, wherein the true value is the theoretical value, the agreed true value under certain measurement conditions is the relative true value, and the relative true value is used instead of the true value; The soil moisture control block construction module constructs the soil moisture control block, and takes the soil moisture control block as the relative true value of the soil moisture standard quantitative product; The authenticity verification module takes the difference between the soil moisture standard quantitative product and the soil moisture control block as the authenticity verification index, which is used for quantitative evaluation of the accuracy and uncertainty of the soil moisture standard quantitative product.

15. An electronic device, comprising: Including: one or more processors; a memory storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to carry out the method of any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that, a memory storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to carry out the method of any one of claims 1 to 7. a memory storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to carry out the method of any one of claims 1 to 7.

Images