Forest region microwave remote sensing earth surface dead combustible moisture content monitoring method and system

Abstract

The invention belongs to the technical field of data processing, and particularly relates to a forest region microwave remote sensing earth surface dead combustible moisture content monitoring method and system, and the method comprises the steps: building a soil background scattering prediction model through historical time sequence data, and obtaining a soil background scattering coefficient in a current environment; constructing a dielectric and energy coupling balance equation considering the two-way attenuation effect, and introducing a moisture impedance attenuation function and a moisture gain scattering function; and constructing a residual target function according to the total backscattering coefficient observed by the satellite and the soil background scattering coefficient, and inverting the moisture content of the dead combustible matters on the earth surface by using a constrained interval search algorithm. According to the method, the soil background and dead combustible signals are decoupled, the problem that a traditional linear addition model loses efficacy in a humid environment is solved, and accurate monitoring of the moisture content of the earth surface dead combustible under the complex hydrothermal condition is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing. More particularly, the present application relates to a forest microwave remote sensing ground surface dead combustible moisture content monitoring method and system. BACKGROUND

[0002] Forest fire is a major natural disaster that destroys forest resources and threatens ecological safety. Ground surface dead combustible moisture content is a key physical parameter for assessing forest fire risk level and predicting fire spread speed. Precise monitoring of the ground surface dead combustible moisture content is of great significance. Traditional field sampling methods are inefficient and cannot meet the needs of large-scale monitoring. L-band microwave remote sensing technology can penetrate the canopy layer to obtain ground surface information due to its strong penetration ability, and thus is applied to ground parameter inversion. Currently, it has become a research hotspot to extract key environmental factors from radar echoes using massive remote sensing data combined with big data analysis techniques, aiming to realize quantitative perception of complex forest environments.

[0003] Existing monitoring methods mainly include physical model simulation, semi-empirical model inversion, and data-driven methods. Among them, semi-empirical models based on linear superposition assumptions are most widely used. Such methods consider that the total backscatter energy received by the satellite is equal to the simple addition of soil scattering energy and combustible scattering energy. Meanwhile, some studies attempt to use physical models to analyze microwave transmission mechanisms for forward simulation, or use statistical analysis techniques to directly establish a mapping relationship between remote sensing data and ground measured data, trying to solve the moisture content inversion problem through data-driven methods.

[0004] However, the above existing technologies have significant limitations. First, the linear superposition assumption seriously ignores the double-path attenuation effect of the dead combustible layer when the moisture content increases. Wet dead combustible layer not only enhances its own scattering, but also significantly absorbs and attenuates the reflected signal from the underlying soil like a water curtain, resulting in a large deviation between the observed signal and the linear model prediction value in wet scenarios. Second, physical models require accurate micro parameters such as ground roughness, which are difficult to obtain at a macro scale, resulting in large errors in soil background signal simulation. SUMMARY

[0005] To solve the technical problems of the above existing technologies ignoring the double-path attenuation effect and the difficulty in obtaining micro parameters, resulting in large deviations in wet scenarios and large errors in background signal simulation, the present application provides solutions in the following aspects.

[0006] In a first aspect, the present application provides a forest microwave remote sensing ground surface dead combustible moisture content monitoring method, comprising: Historical microwave remote sensing data and meteorological data of the target forest area were acquired, and a sample set of approximately bare soil was selected. A soil background scattering prediction model was constructed using machine learning to predict the soil background scattering coefficient under current environmental conditions. A dielectric and energy coupling balance equation that takes into account the two-way attenuation effect was constructed. The dielectric and energy coupling balance equation includes a moisture impedance attenuation function and a moisture gain scattering function. The dielectric and energy coupling balance equation is used to characterize the nonlinear relationship between the total backscattering coefficient, the soil background scattering coefficient, and the surface dead combustible water content. Based on the total backscattering coefficient actually observed by the satellite sensor and the soil background scattering coefficient, a residual objective function was constructed. The dielectric and energy coupling balance equation was solved using an interval search algorithm to obtain the surface dead combustible water content.

[0007] This invention utilizes historical time-series data to screen an approximate bare soil sample set and combines it with machine learning to construct a soil background scattering prediction model, providing a zero potential energy surface benchmark for subsequent inversion. Furthermore, by constructing a dielectric-energy coupling balance equation that includes a moisture impedance attenuation function and a moisture gain scattering function, this invention physically measures the nonlinear contribution of dead combustible material moisture content changes to the total scattering signal, solving the problem of traditional linear additive models failing in humid environments and avoiding estimation errors caused by excessive subtraction of soil signals or neglect of shading effects. Finally, this invention constructs a residual objective function based on the total backscattering coefficient and the soil background scattering coefficient and solves it using an interval search algorithm, effectively addressing the problems of multiple solutions and non-monotonicity in the solution of transcendental equations, ensuring the physical authenticity of the inversion results and avoiding getting trapped in local optima, thereby achieving accurate monitoring of the moisture content of dead combustible material on the forest surface.

[0008] Preferably, the screening of the approximate bare soil sample set includes: calculating the normalized vegetation index based on optical remote sensing data; selecting pixels in the historical time series whose normalized vegetation index is less than the dynamic vegetation threshold as approximate bare soil samples; removing data within a preset time period after rainfall; and using the remaining data as the approximate bare soil sample set.

[0009] This invention effectively identifies approximately bare soil pixels in historical time series by calculating the normalized vegetation index based on optical remote sensing data and combining it with dynamic vegetation thresholds for screening. At the same time, by removing data within a preset time period after rainfall, it eliminates the interference of temporary surface water accumulation or moist vegetation on soil scattering characteristics, ensuring that the sample set can truly reflect the scattering characteristics of pure soil under different hydrothermal conditions. This provides a high-quality and representative data foundation for the subsequent construction of a high-precision soil background scattering prediction model.

[0010] Preferably, the step of constructing a soil background scattering prediction model using machine learning includes: constructing a soil background scattering prediction model using an XGBoost model, wherein the input features of the soil background scattering prediction model include soil volumetric water content, surface temperature, the square of soil volumetric water content, and monthly features, the labels are the backscattering coefficients corresponding to microwave remote sensing data in a selected set of approximate bare soil samples, the loss function is set as the mean square error loss function, and the soil background scattering prediction model is trained using the approximate bare soil sample set.

[0011] This invention uses the XGBoost model to construct a soil background scattering prediction model and selects key meteorological factors such as soil volumetric water content and surface temperature as input features. It makes full use of the powerful nonlinear fitting ability of machine learning, which can more accurately predict the soil background scattering coefficient under the current environmental conditions. This provides a zero potential energy surface reference that removes the interference of vegetation and dead combustibles for subsequent physical stripping, and avoids the deviation in the inversion of dead combustible water content caused by the reference error.

[0012] Preferably, the dielectric and energy coupling balance equation satisfies the following expression: In the formula, This represents the total backscattering coefficient actually observed by the satellite sensor; This represents the soil background scattering coefficient predicted by the soil background scattering prediction model; This represents the moisture impedance attenuation function; This represents the moisture gain scattering function.

[0013] This invention constructs a dielectric and energy coupling balance equation, organically combining the moisture impedance attenuation function and the moisture gain scattering function to measure the nonlinear contribution of dead combustible material moisture content changes to the total scattering signal at the physical level. By modeling the dead combustible material moisture content as both a scattering source and an attenuation barrier, it effectively solves the problem of the failure of the traditional linear additive model in humid environments. It avoids moisture content estimation errors caused by excessive subtraction of soil signals or neglect of shading effects when dead combustible material is wet, thus improving the applicability of the model under complex hydrothermal conditions.

[0014] Preferably, the moisture impedance attenuation function satisfies the expression: In the formula, This represents the moisture impedance attenuation function; Represents an exponential function with the natural constant as its base; Indicates the optical thickness coefficient of the combustible layer; This indicates the moisture content of dead combustible material on the surface to be inverted; The threshold for saturated moisture content of dead combustible material; Represents the structural shape factor; Indicates the angle of incidence for satellite observation. This represents the cosine function.

[0015] The moisture impedance attenuation function of this invention uses an exponential function with the natural constant as the base to characterize the attenuation law of electromagnetic waves when penetrating a lossy medium layer. It comprehensively considers the optical thickness coefficient of the combustible layer, the structural shape factor, and the correction effect of the satellite observation incident angle on the path length. It can accurately measure the energy residual ratio of microwave signals after penetrating the dead combustible layer, and measure the bidirectional shielding effect of the moist combustible layer on the signal of the underlying soil, ensuring the accuracy of the calculation of the contribution of soil background signal.

[0016] Preferably, the moisture gain scattering function satisfies the expression: In the formula, Represents the moisture gain scattering function; Represents an exponential function with the natural constant as its base; Indicates the optical thickness coefficient of the combustible layer; This indicates the moisture content of dead combustible material on the surface to be inverted; The threshold for saturated moisture content of dead combustible material; Represents the structural shape factor; Indicates single-scatter albedo; Indicates the angle of incidence for satellite observation. This represents the cosine function.

[0017] This invention introduces a moisture gain scattering function, which, based on the energy conservation approximation, sets the volume scattering intensity to be proportional to the energy intercepted by the medium layer. The proportion of the scattering signal converted into a single scattering albedo is controlled by the single scattering albedo, and the volume scattering term is geometrically normalized using a cosine function. This allows for the measurement of the contribution of the backscattering energy generated by the dead combustible material layer itself to the total signal, thus achieving the characterization of the volume scattering effect of the dead combustible material layer.

[0018] Preferably, the residual objective function satisfies the expression: In the formula, This represents the residual objective function value; Indicates the absolute value symbol; This represents the total backscattering coefficient actually observed by the satellite sensor; This represents the soil background scattering coefficient predicted by the soil background scattering prediction model; Represents an exponential function with the natural constant as its base; Indicates the optical thickness coefficient of the combustible layer; This indicates the moisture content of dead combustible material on the surface to be inverted; The threshold for saturated moisture content of dead combustible material; Represents the structural shape factor; Indicates single-scatter albedo; This indicates the angle of incidence for satellite observation.

[0019] Preferably, solving the dielectric-energy coupling balance equation using the interval search algorithm includes: in response to the total backscattering coefficient being less than the soil background scattering coefficient, setting the lower bound of the search interval to... The upper limit is set as In response to the total backscattering coefficient being greater than the soil background scattering coefficient, the lower bound of the search interval is set to 0, and the upper bound is set to... Within the search interval, iteratively search using the golden section method until the search interval length is less than the preset tolerance error, outputting the surface dead combustible water content that minimizes the residual objective function value; in response to the total backscattering coefficient being equal to the soil background scattering coefficient, [the following is done / is done / etc.]. As the inversion result of the surface dead combustible material moisture content; among which, This represents the threshold of saturated moisture content of dead combustible materials.

[0020] This invention dynamically adjusts the search center by judging the relative magnitude of the total backscattering coefficient and the soil background scattering coefficient. When the total backscattering coefficient is less than the soil background scattering coefficient, it is determined that the attenuation effect is dominant, and the high water content range is locked; otherwise, it is determined that the scattering enhancement effect is dominant, and the low water content range is locked. This not only solves the problems of multiple solutions and non-monotonicity in the solution of transcendental equations, but also ensures the convergence speed of the inversion algorithm, ensures the physical authenticity of the inversion results, and avoids the mathematical solution from getting trapped in local optima.

[0021] Preferably, it further includes: in response to the residual objective function value after iterative convergence exceeding a preset tolerance threshold, marking the current pixel as an invalid value, and replacing it with the interpolation result of neighboring pixels.

[0022] Secondly, the present invention provides a microwave remote sensing system for monitoring the moisture content of dead combustibles on the surface of forest areas, comprising a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the aforementioned method for monitoring the moisture content of dead combustibles on the surface of forest areas using microwave remote sensing is implemented.

[0023] By adopting the above technical solution, a computer program is generated from the above-mentioned method for monitoring the moisture content of dead combustibles on the surface of forest areas using microwave remote sensing, and stored in a memory so that it can be loaded and executed by a processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0024] The beneficial effects of this invention are as follows: This invention employs a strategy combining spatiotemporal filtering and machine learning to construct a soil background scattering prediction model. The model is trained using data from historical bare soil windows, enabling it to learn the nonlinear scattering characteristics of soil under specific meteorological conditions. This provides a zero potential energy surface reference that removes interference from vegetation and dead combustibles for subsequent physical stripping, thereby avoiding the deviation in the inversion of dead combustible moisture content caused by reference errors.

[0025] This invention constructs a dielectric and energy coupling balance equation that takes into account the two-way attenuation effect, and introduces a moisture impedance attenuation function and a moisture gain scattering function to measure the nonlinear contribution of dead combustible material moisture content change to the total scattering signal at the physical level. This invention considers that the subtracted soil background should not be a simple bare soil background, but a soil background attenuated by the dead combustible material layer. The dead combustible material moisture content is modeled as both a scattering source and an attenuation barrier, avoiding moisture content estimation errors caused by excessive subtraction of soil signals or neglect of the shading effect when the dead combustible material is wet. It effectively solves the problem of the failure of the traditional linear additive model in humid environments and improves the monitoring robustness under complex hydrothermal conditions.

[0026] This invention designs a specific residual objective function and uses a constrained interval search algorithm for numerical inversion. By judging the relative magnitude of the observed value and the predicted soil background value, the search center is dynamically adjusted, effectively solving the problems of multiple solutions and non-monotonicity in the solution of transcendental equations. When the attenuation effect is dominant, the focus is on the high moisture content interval, and when the scattering enhancement effect is dominant, the focus is on the medium and low moisture content interval. This not only ensures the convergence speed of the inversion algorithm, but also ensures the physical authenticity of the inversion results, avoiding the mathematical solution from getting trapped in local optima. It realizes high-precision and automated monitoring of the moisture content of dead combustibles on the forest surface. Attached Figure Description

[0027] Figure 1 This is a schematic flowchart illustrating a method for monitoring the moisture content of dead combustibles on the surface of forest areas using microwave remote sensing, according to the present invention. Figure 2 This is a time-series comparison of the total backscattering coefficient and the soil background scattering coefficient; Figure 3 A time-series comparison of the inversion results of surface dead combustible water content based on the dielectric and energy coupling balance equation with the actual value; Figure 4 Scatter plot to verify the accuracy of the inversion. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0030] This invention discloses a method for monitoring the moisture content of dead combustibles on the surface of forest areas using microwave remote sensing, with reference to... Figure 1 This includes steps S1-S3: S1. Obtain historical time-series microwave remote sensing data and meteorological data of the target forest area, screen a sample set of approximately bare soil, and use machine learning to construct a soil background scattering prediction model to predict the soil background scattering coefficient under the current environmental conditions.

[0031] It should be noted that, in order to extract information about dead combustibles from the total signal received from satellites, the primary task is to determine the soil background signal under conditions without combustible obstruction. However, soil scattering intensity is nonlinearly affected by various factors such as soil moisture, soil texture, and surface roughness, making it difficult to accurately simulate the soil background signal on a macroscopic scale using traditional physical models. Therefore, this invention employs a strategy combining spatiotemporal filtering and machine learning, utilizing the bare soil window periods existing in historical time-series data to train the model to learn the scattering characteristics of soil under specific meteorological conditions, thereby providing a benchmark for subsequent physical stripping.

[0032] Specifically, historical long-term L-band microwave remote sensing data of the target forest area is acquired. This L-band microwave remote sensing data consists of regularly arranged pixels. The backscattering coefficient of each pixel is extracted, and radiometric calibration and topographic radiometric correction are performed on the L-band microwave remote sensing data. Concurrent reanalysis meteorological data is acquired, including shallow soil volumetric water content, surface temperature, and cumulative rainfall. The spatial resolution of the meteorological data is resampled to the same spatial resolution as the microwave remote sensing data using bicubic interpolation, and a time window is matched.

[0033] Furthermore, the normalized vegetation index (NDI) is calculated using concurrent optical remote sensing data. The NDI is determined using the absorption characteristics of vegetation to red light and its reflectance characteristics to near-infrared light, satisfying the expression: ; In the formula, Indicates the normalized vegetation index; Indicates the reflectivity in the near-infrared band; This indicates the reflectivity in the red light band.

[0034] A dynamic vegetation threshold is set, and pixels with a normalized vegetation index (NDI) less than the threshold in historical time series are selected as approximate bare soil samples or those with sparse vegetation cover. It should be noted that... Typically, for areas such as bare soil and rocks, selecting 0.1 as the dynamic vegetation threshold can effectively filter out bare soil pixels. Therefore, in this embodiment, the dynamic vegetation threshold is set to 0.1. In other embodiments, implementers can set the dynamic vegetation threshold according to the vegetation sparseness of the target forest area.

[0035] To avoid the influence of high-moisture dead combustibles that have not dried after rainfall on the samples, a rainfall event filtering method is used to remove data within a preset time period after the rainfall occurs. This ensures that the moisture in the samples mainly comes from the soil rather than the surface cover, and the filtered data is used as an approximate bare soil sample set. It should be noted that the water film on the surface of the land and vegetation after rainfall usually evaporates or infiltrates within one day. Retaining a 24-hour buffer period can effectively avoid the interference of the surface water film on the dielectric constant. Therefore, in this embodiment, the preset time period is set to 24 hours. In other embodiments, the implementer can set the preset time period within the range of 12 to 48 hours according to the local climate drying rate.

[0036] A soil background scattering prediction model is constructed. In this embodiment, the soil background scattering prediction model adopts the Extreme Gradient Boosting (XGBoost) model. XGBoost is a high-efficiency ensemble learning algorithm whose core base learner is a decision tree. This model can effectively capture the complex nonlinear relationship between input features and output labels by integrating the prediction results of multiple decision trees. The input features of the soil background scattering prediction model include soil volumetric water content, surface temperature, the square of soil volumetric water content, and monthly features. The labels are the backscattering coefficients corresponding to microwave remote sensing data in the selected set of approximate bare soil samples. The loss function is set as the mean squared error loss function. The soil background scattering prediction model is trained using the approximate bare soil sample set. In other embodiments, implementers can also use other machine learning algorithms to construct the soil background scattering prediction model, such as the Random Forest Regression (RFR) model or the Support Vector Regression (SVR) model.

[0037] During the real-time monitoring phase, the current meteorological data is input into the trained soil background scattering prediction model, and the soil background scattering coefficient under the current environmental conditions is output.

[0038] For example, Figure 2The graph shows a time-series comparison between the total backscattering coefficient actually observed by the satellite sensor and the soil background scattering coefficient predicted by the soil background scattering prediction model. The non-overlapping parts of the two curves intuitively reflect the attenuation effect and volume scattering contribution of the dead combustible layer on the surface to the microwave signal.

[0039] S2. Construct a dielectric and energy coupling balance equation that takes into account the two-way attenuation effect. The dielectric and energy coupling balance equation includes a moisture impedance attenuation function and a moisture gain scattering function. The dielectric and energy coupling balance equation is used to characterize the nonlinear relationship between the total backscattering coefficient, the soil background scattering coefficient and the surface dead combustible water content.

[0040] It should be noted that the microwave signal undergoes two-way attenuation—one incident attenuation and one reflection attenuation—as it penetrates the dead combustible material layer, reaches the soil, and returns to the sensor. Changes in the moisture content of the dead combustible material simultaneously alter its own volumetric scattering characteristics and its transmission characteristics to the underlying soil signal. Higher moisture content results in lower transmittance and a stronger masking effect on the soil signal. Furthermore, the increased volumetric scattering from the dead combustible material layer means that traditional additive models neglect the dynamic process of transmittance changing with moisture content, leading to inversion failure under humid conditions. Therefore, this invention constructs a coupled equation that models the moisture content of the dead combustible material as both a scattering source and an attenuation barrier, thereby physically reconstructing the true signal transmission process.

[0041] Specifically, the dielectric and energy coupling balance equation satisfies the following expression: ; In the formula, This represents the total backscattering coefficient actually observed by the satellite sensor; This represents the soil background scattering coefficient predicted by the soil background scattering prediction model; This represents the moisture impedance attenuation function, used to characterize the bidirectional shielding effect of the moist combustible layer on signals from the underlying soil. Its physical meaning is the energy residue rate of a microwave signal after it has passed through the dead combustible layer. As the moisture content of the dead combustible increases... It decreases exponentially, leading to a decrease in the soil background scattering coefficient. For the total backscattering coefficient The contribution decreased; This represents the moisture gain scattering function, used to characterize the volumetric scattering energy generated by the moist combustible layer itself. As the moisture content of the dead combustible increases... The increase leads to an increase in its total backscattering coefficient. The contribution increased.

[0042] Furthermore, the moisture impedance attenuation function and the moisture gain scattering function satisfy the following expression: ; ; In the formula, This represents the moisture impedance attenuation function; Represents the moisture gain scattering function; Represents an exponential function with the natural constant as its base; The optical thickness coefficient of the combustible layer is a dimensionless parameter that characterizes the energy loss rate of microwaves penetrating the dead combustible layer. It is determined by the forest stand type; when the forest stand type is coniferous forest... The value ranges from 0.3 to 0.5 when the forest stand type is broadleaf forest. The value range is from 0.1 to 0.3. The larger the value, the greater the loss during microwave penetration. Implementers can set the optical thickness coefficient according to the actual implementation situation. For example, it can be set to 0.4 for coniferous forests and 0.2 for broad-leaved forests. This indicates the moisture content of dead combustible material on the surface to be inverted; This represents the saturated moisture content threshold of dead combustibles, used to normalize the moisture content. The empirical range is 250% to 300%. Implementers can set the saturated moisture content threshold according to the actual implementation situation, for example, a value of 300%. This represents the structural shape factor, used to correct for non-Lambertian decay characteristics caused by non-uniform media. When the dead combustible material is a flat, fallen leaf... The value is 1.0 when the dead combustible material is a cylindrical dead branch. The value ranges from 1.2 to 1.5. In this embodiment, when the dead combustible material is flat fallen leaves, The value is 1.0 when the dead combustible material is a cylindrical dead branch. The value is set to 1.2. In other embodiments, implementers can set the structural shape factor according to the actual implementation situation. The larger the value, the higher the nonlinearity of the attenuation characteristics; This represents the single-scattering albedo, which reflects the ability of the dead combustible layer to convert incident energy into volumetric scattered energy. It is determined by the microscopic geometry and dielectric properties of the dead combustible. Implementers can set this value based on the main vegetation type of the target forest area and prior knowledge of the microwave radiative transfer model. For example, in the L-band... The typical value range is 0.06 to 0.12, and in this embodiment, the value is 0.08; Indicates the angle of incidence for satellite observation. Represents the cosine function, with the exponential term. In The moisture gain scattering function is used to correct the vertical thickness to the actual tilt path length of microwave transmission. The product term in Used for geometric normalization of volume scattering terms.

[0043] It should be noted that this invention follows Beer-Lambert's law in microwave transmission theory, using an exponential function to characterize the attenuation of electromagnetic waves penetrating a lossy dielectric layer, and constructs a moisture impedance attenuation function. Considering that the dead combustible material layer is a non-homogeneous medium and that the effect of water content on the dielectric constant has non-linear characteristics, an optical thickness coefficient is used. With structural shape factor We jointly constructed a nonlinear optical depth model and utilized... The vertical thickness is corrected to the actual inclined path length of microwave transmission, thereby accurately measuring the proportion of energy remaining after microwaves pass through the dead combustible material layer.

[0044] This invention, based on the energy conservation approximation, assumes that the volumetric scattering intensity is proportional to the energy intercepted by the medium layer, and constructs a moisture gain scattering function. The intercepted energy portion is determined by the single-scattering albedo. Control the proportion converted into a scattered signal, and use the product term The volume scattering term is geometrically normalized to measure the contribution of the backscattered energy generated by the dead combustible layer itself to the total signal.

[0045] S3. Based on the total backscattering coefficient actually observed by the satellite sensor and the soil background scattering coefficient, construct the residual objective function, and use the interval search algorithm to solve the dielectric and energy coupling balance equation to obtain the surface dead combustible water content.

[0046] It should be noted that, since the dielectric-energy coupling balance equation is a transcendental equation concerning water content, it is highly nonlinear and non-monotonic, and cannot be solved analytically directly. Furthermore, because the contributions of the water impedance attenuation function and the water gain scattering function to the total energy show opposite trends, the total signal may exhibit non-monotonicity as water content changes. Therefore, this invention designs a specific residual objective function and uses the golden section search method to lock in the true water content value within the physically constrained interval, thereby solving the problem of multiple solutions in the inversion.

[0047] Specifically, the residual objective function satisfies the expression: ; In the formula, The residual objective function value represents the difference between the observed values ​​and the theoretical model values. The smaller the value, the closer the retrieved moisture content is to the true value; Indicates the absolute value symbol; This represents the total backscattering coefficient actually observed by the satellite sensor; This represents the soil background scattering coefficient predicted by the soil background scattering prediction model; Represents an exponential function with the natural constant as its base; Indicates the optical thickness coefficient of the combustible layer; This indicates the moisture content of dead combustible material on the surface to be inverted; The threshold for saturated moisture content of dead combustible material; Represents the structural shape factor; Indicates single-scatter albedo; Indicates the angle of incidence for satellite observation. This represents the cosine value of the angle of incidence.

[0048] The physical range of the surface dead combustible material moisture content is set to 0 to... The total backscattering coefficient actually observed by the satellite sensor. The soil background scattering coefficient is less than that predicted by the soil background scattering prediction model. The determination that the attenuation effect is dominant indicates that the moist dead combustible layer has a strong masking effect on the soil signal. Therefore, the lower bound of the search interval is set accordingly. Set as Upper Realm Set as ; Response to the total backscattering coefficient actually observed by satellite sensors Greater than the soil background scattering coefficient predicted by the soil background scattering prediction model The determination that the scattering enhancement effect is dominant indicates that volume scattering contributes significantly while the shading effect is relatively weak. At this point, the dead combustible material may be in a low moisture content state or the soil background may be extremely dry, thus lowering the search interval. Set to 0, upper bound Set as ; Response to the total backscattering coefficient actually observed by satellite sensors Equal to the soil background scattering coefficient predicted by the soil background scattering prediction model This indicates that the dead combustible material layer is in a specific intermediate equilibrium state, at which point directly applying... As an inversion result, there is no need to determine the search interval for iterative search.

[0049] Determine the lower bound of the search interval. With the upper realm Then, within this search interval, two trial points are initialized using the golden ratio, and the two trial points satisfy the expression: ; ; In the formula, This indicates the first tentative step; This indicates the second probing point; Indicates the lower bound of the search interval; This indicates the upper bound of the search interval.

[0050] Calculate the residual objective function values ​​corresponding to the two trial points, and compare the magnitudes of the residual objective function values ​​to update the lower bound of the search interval. or upper boundary The test points are recalculated within the updated search interval, and this process is iterated until the search interval length is less than a preset tolerance error. The surface dead combustible water content that minimizes the residual objective function value is output as the final inversion result. In this embodiment, the tolerance error is set to 0.01, meaning that iteration stops when the interval length is less than 0.01. At this point, the numerical accuracy of the inversion result meets the monitoring requirements. In other embodiments, the implementer can set the tolerance error according to the actual requirements for inversion accuracy and computational efficiency.

[0051] Furthermore, in response to the residual objective function value after iterative convergence still exceeding the preset tolerance threshold, it is determined that the physical model cannot explain the current observation, the pixel is marked as invalid, and the interpolation result based on the neighboring pixels is output as a substitute.

[0052] This achievement enables the monitoring of moisture content in dead combustibles on the forest surface using microwave remote sensing.

[0053] For example, Figure 3 The image shows a time-series comparison between the inversion results of surface dead combustible moisture content based on the dielectric and energy coupling balance equation and the actual value. The inversion results closely follow the fluctuations of the actual value throughout the time series. In particular, no significant deviations were observed in the high and low moisture content ranges, indicating that the dielectric and energy coupling balance equation and dynamic range search strategy of the present invention can effectively decouple the soil background and dead combustible signal. Figure 4 To verify the accuracy of the inversion, a scatter plot was used. The distribution of the monitoring pixels closely surrounded the 1:1 ideal reference line, with a mean square error of 0.0521. This verified the high accuracy and robustness of the invention in monitoring the moisture content of dead combustibles on the ground under different environmental conditions.

[0054] This invention also discloses a microwave remote sensing system for monitoring the moisture content of dead combustibles on the surface of forest areas, comprising a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a method for monitoring the moisture content of dead combustibles on the surface of forest areas according to the present invention.

[0055] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A method for monitoring the moisture content of ground dead combustible material in a forest area by means of microwave remote sensing, characterized in that The method comprises the following steps: acquiring microwave remote sensing data and meteorological data of a target forest area at historical time points, screening an approximate bare soil sample set, and constructing a soil background scattering prediction model by using machine learning to predict the soil background scattering coefficient under current environmental conditions; constructing a dielectric and energy coupling balance equation considering the two-way attenuation effect, the dielectric and energy coupling balance equation comprising a water impedance attenuation function and a water gain scattering function, and using the dielectric and energy coupling balance equation to represent the nonlinear relationship between the total backscattering coefficient, the soil background scattering coefficient and the surface dead combustible moisture content; constructing a residual target function according to the total backscattering coefficient and the soil background scattering coefficient actually observed by a satellite sensor, and solving the dielectric and energy coupling balance equation by using an interval search algorithm to obtain the surface dead combustible moisture content.

2. The method according to claim 1, wherein the method is characterized by, The screening of the approximate bare soil sample set comprises the following steps: calculating a normalized vegetation index according to optical remote sensing data, screening out pixels with a normalized vegetation index less than a dynamic vegetation threshold value in the historical time points as approximate bare soil samples, and removing data within a preset time period after rainfall to obtain the approximate bare soil sample set.

3. The method according to claim 1, wherein, The construction of the soil background scattering prediction model by using machine learning comprises the following steps: constructing a soil background scattering prediction model by using an XGBoost model, the input features of the soil background scattering prediction model comprising soil volume water content, surface temperature, square of the soil volume water content and month features, the label being the backscattering coefficient corresponding to the microwave remote sensing data in the screened approximate bare soil sample set, the loss function being set as a mean square error loss function, and the approximate bare soil sample set being used to train the soil background scattering prediction model.

4. The method according to claim 1, wherein, The dielectric and energy coupling balance equation satisfies the expression: ； wherein represents the total backscatter coefficient actually observed by the satellite sensor; represents the soil background backscatter coefficient predicted by the soil background scattering prediction model; represents the water retardation attenuation function; represents the water gain scattering function.

5. The method according to claim 4, wherein the method is characterized by, The water impedance attenuation function satisfies the expression: ； wherein, represents a moisture impedance decay function; represents an exponential function with a natural constant as base; represents an optical thickness coefficient of the combustible layer; represents the surface dead combustible moisture content to be inverted; represents a saturated moisture content threshold of the dead combustible; represents a structure shape factor; represents an incidence angle of the satellite observation, represents a cosine function.

6. The method according to claim 4, wherein the method is characterized by, The water gain scattering function satisfies the expression: ； wherein denotes the water gain scattering function; denotes the exponential function with base of the natural constant; denotes the optical thickness coefficient of the combustible layer; denotes the surface dead combustible water content to be inverted; denotes the saturation water content threshold of the dead combustible; denotes the structure shape factor; denotes the single scattering albedo; denotes the satellite observation incidence angle, denotes the cosine function.

7. The method according to claim 1, wherein the method is characterized by, The residual target function satisfies the expression: ； wherein, denotes the residual objective function value; denotes the absolute value sign; denotes the total backscatter coefficient actually observed by the satellite sensor; denotes the soil background backscatter coefficient predicted by the soil background scattering prediction model; denotes the exponential function with base of the natural constant; denotes the optical thickness coefficient of the combustible layer; denotes the surface dead combustible moisture content to be inverted; denotes the saturated moisture content threshold of the dead combustible; denotes the structure shape factor; denotes the single scattering albedo; denotes the satellite observed incidence angle.

8. The method according to claim 1, wherein the method is characterized by, The solving of the dielectric and energy coupling balance equation by using the interval search algorithm comprises the following steps: in response to the total backscattering coefficient being less than the soil background scattering coefficient, setting the lower bound of the search interval as and the upper bound as ; in response to the total backscattering coefficient being greater than the soil background scattering coefficient, setting the lower bound of the search interval as 0 and the upper bound as ; iteratively searching in the search interval by using the golden section method until the length of the search interval is less than a preset tolerance error, and outputting the surface dead combustible moisture content that minimizes the residual objective function value; in response to the total backscattering coefficient being equal to the soil background scattering coefficient, taking as the inversion result of the surface dead combustible moisture content; wherein represents the saturated moisture content threshold of the dead combustible.

9. The method according to claim 8, wherein the method is characterized by, The method further comprises the following steps: in response to the residual target function value after iteration convergence exceeding a preset tolerance threshold, marking the current pixel as an invalid value, and replacing it with an interpolation result of neighboring pixels.

10. A forested area microwave remote sensing system for monitoring the moisture content of surface dead combustible material, comprising: The method comprises the following steps: a processor and a memory, the memory storing computer program instructions, when the computer program instructions are executed by the processor, a forest area microwave remote sensing surface dead combustible moisture content monitoring method according to any one of claims 1-9 is realized.

Images