Multi-parameter GNSS-R soil humidity inversion method based on random forest model

Abstract

The invention provides a multi-parameter GNSS-R soil humidity inversion method based on a random forest model. The multi-parameter GNSS-R soil humidity inversion method comprises the following steps: satellite-borne GNSS-R data quality control; satellite-borne GNSS-R data observation quantity extraction and surface reflectance calculation are carried out; eliminating satellite-borne GNSS-R observation data influenced by the open water area; carrying out surface reflectance gridding; preprocessing the SMAP data; constructing an inversion model; and carrying out spaceborne GNSS-R soil humidity inversion and precision evaluation. The method has the beneficial effects that by utilizing the performance advantages of processing a non-linear relationship, good generalization performance, strong robustness and the like of the random forest, after observation data influenced by a water body is accurately eliminated, an inversion model is constructed by combining multi-parameter auxiliary data of a soil humidity active and passive remote sensing satellite; the space-time coverage, the inversion precision and the reliability of soil humidity data are remarkably improved, and powerful data support is provided for the fields of climate forecast, agricultural production, hydrological modeling and the like.

Description

Technical Field

[0001] The present invention belongs to the field of soil moisture inversion, and in particular relates to a multi-parameter GNSS-R soil moisture inversion method based on a random forest model. Background Art

[0002] Soil moisture is a key parameter in the process of energy exchange between land and atmosphere. As an important part of the global hydrological cycle, its distribution and spatial variation characteristics have a profound impact on the hydrological and meteorological processes in the world or local areas, which is conducive to more accurate hydrological modeling and support for social decision-making. Traditional soil moisture measurement technology is time-consuming and labor-intensive, and cannot achieve large-scale monitoring. Although commonly used remote sensing methods such as optical remote sensing and microwave remote sensing make up for the defects of direct observation, they also have many problems such as high satellite cost, high power consumption, limited temporal and spatial resolution, and difficulty in networking and formation. Global Navigation Satellite System-Reflectometry (GNSS-R) uses many GNSS satellites as signal sources. The receiver can be flexibly mounted on various platforms. By receiving and processing the reflected signals from the surface, the inversion of the target geophysical parameters is achieved. Satellite-borne GNSS-R technology has the advantages of low cost, low power consumption, short replay period, and no weather influence, and has become an important technical means for soil moisture monitoring. There is an obvious coupling relationship between surface reflectivity and soil moisture. Many GNSS-R soil moisture inversion models have been developed using dielectric constant as an intermediate parameter. However, since surface reflectivity is also affected by other surface parameters such as roughness, vegetation, and water bodies, the existing empirical models have problems such as limited applicable areas and low inversion accuracy. Summary of the invention

[0003] In view of this, the present invention aims to propose a multi-parameter GNSS-R soil moisture inversion method based on a random forest model to solve the problem of high-precision real-time monitoring of soil moisture. This problem may lead to a decrease in the accuracy of decision-making in fields such as agricultural irrigation, water resources management and weather forecasting. Solving this problem is conducive to improving the accuracy and reliability of soil moisture monitoring, and providing strong support for scientific research, production practice and disaster warning in the fields of agriculture, hydrology, meteorology, etc.

[0004] To achieve the above object, the technical solution of the present invention is achieved as follows: A multi-parameter GNSS-R soil moisture inversion method based on a random forest model includes the following steps: S1. Perform quality control on onboard GNSS-R data; S2, extracting the observed quantity of the satellite-borne GNSS-R data after quality control in step S1 and calculating the surface reflectivity to obtain the surface reflectivity, and proceeding to step S4; S3, eliminating the satellite-borne GNSS-R observation data affected by open waters, and proceeding to step S4; S4, gridding the surface reflectivity obtained in step S2 using the spaceborne GNSS-R observation data obtained in step S3 to obtain gridded data; S5, performing SMAP data preprocessing in combination with the gridded data in step S4; S6, constructing an inversion model; S7. Spaceborne GNSS-R soil moisture inversion and accuracy assessment.

[0005] Furthermore, in step S1, the quality control of the spaceborne GNSS-R data includes: Quality control is done by both quality labeling and empirical guidelines; First, the quality label data is converted into binary, and the data quality is determined based on the bit value. The label bits used include blackbody load delay Doppler pattern mode, radio frequency interference, and test mode. Secondly, the data was screened again according to empirical criteria, and the variables and threshold conditions used included: The incident angle is greater than 65°, the signal-to-noise ratio is less than 0, and the antenna gain is less than 0; Filter out the data that meets the conditions.

[0006] Furthermore, in step S2, the observation quantity of the satellite-borne GNSS-R data after quality control in step S1 is extracted and the surface reflectivity is calculated to obtain the surface reflectivity, including: The latitude and longitude coordinates of the mirror point of the satellite-borne GNSS-R data and other observations used to calculate the surface reflectance are extracted. The surface reflectivity is calculated based on the bistatic radar equation. Assuming that coherent scattering occurs on land, the coherent component is expressed in the following form: (1) In the formula, is the coherent scattering component power; r represents the right-hand circularly polarized antenna of the GNSS transmitter, and l represents the left-hand circularly polarized antenna of the GNSS-R receiver; is the signal wavelength; is the peak power of the transmitted signal; , are the gains of the transmitting antenna and the receiving antenna respectively; , are the distances between the transmitter, receiver and the mirror reflection point respectively; represents the surface reflectivity; The power expression of the incoherent scattered component is: (2) In the formula, is the incoherent scattering component power; Bistatic Radar Cross Section (BRCS), in m 2 ; make = , combine formula (1) and formula (2) to get the surface reflectivity writing Function, such as formula (3): (3).

[0007] Furthermore, in step S3, the satellite-borne GNSS-R observation data affected by open waters is eliminated, including: The seasonal product of the global water dataset is used to mask the water area to obtain mask data, and the open water area is removed by coordinate matching based on the mask data; 0 represents the anhydrous state, and 1 to 12 are marked with 1 representing the water state; In case of missing data, a null value is assigned; Based on the longitude and latitude information, the geographical location, area size and water state represented by each matrix element can be obtained, thereby obtaining the water body data used to mask the GNSS-R sampling points; the longitude and latitude of all GNSS-R sampling points in the study area are extracted and compared with the obtained geographical coordinates of the water body data. If the mirror point is located in the area covered by the open water in the mask data, it will be eliminated.

[0008] Further, in step S4, the surface reflectivity obtained in step S2 is gridded by the spaceborne GNSS-R observation data obtained in step S3 to obtain gridded data, including: According to step S3, the GNSS-R data falling in the open water area is eliminated; and the surface reflectivity of the GNSS-R data not falling in the open water area is calculated, and then the surface reflectivity is gridded using the EASE-Grid of the SMAP data, and the surface reflectivity of the sampling points falling in each grid is averaged to represent the surface reflectivity of this grid.

[0009] Furthermore, in step S5, SMAP data preprocessing is performed in combination with the gridded data in step S4, including: Daily SMAP data include soil moisture estimates, quality labels, vegetation optical thickness, land cover types, and grid-related coordinates of the ascending and descending regions. The grid-related coordinates of the ascending and descending regions are averaged to represent the results of one day. To obtain global coverage, data from three consecutive days were averaged, ignoring null values; SMAP grid coordinates, vegetation optical thickness, surface roughness, land cover type, and annual accumulation day are used as auxiliary data for the inversion model; The soil moisture of SMAP is used as the true value of the GNSS-R soil moisture inversion result in step S7; SMAP data were filtered using quality labels.

[0010] Furthermore, in step S6, an inversion model is constructed, including: All daily observation data were randomly divided into two groups, forming a training set and a test set, which accounted for 70% and 30% of the total data, respectively. Ten-fold cross-validation was used to tune the hyperparameters of the random forest model.

[0011] Furthermore, in step S7, the spaceborne GNSS-R soil moisture inversion and accuracy assessment includes: The input features are substituted into the inversion model to obtain the spaceborne GNSS-R soil moisture inversion results, and the quality of the results is evaluated from three perspectives: model accuracy, temporal variability, and spatial variability.

[0012] Compared with the prior art, the multi-parameter GNSS-R soil moisture inversion method based on the random forest model described in the present invention has the following advantages: The present invention discloses a multi-parameter GNSS-R soil moisture inversion method based on a random forest model. The method utilizes the performance advantages of random forest, such as being able to handle nonlinear relationships, having good generalization performance, and having strong robustness. After accurately eliminating the observation data affected by water bodies, the inversion model is constructed by combining the multi-parameter auxiliary data of the Soil Moisture Active Passive (SMAP) satellite. The method can significantly improve the spatiotemporal coverage, inversion accuracy, and reliability of soil moisture data, and provide strong data support for climate forecasting, agricultural production, hydrological modeling, and other fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings: Figure 1 It is a schematic diagram of the inversion method flow according to an embodiment of the present invention; Figure 2 This is a schematic diagram of an example of seasonal products of GSW according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the distribution of sampling points of a single satellite and a single day of the satellite-borne GNSS-R according to an embodiment of the present invention; Figure 4 A schematic diagram of the scattered distribution of the GNSS-R inversion results and SMAP reference data according to an embodiment of the present invention; Figure 5 A schematic diagram of a random grid time series of GNSS-R inversion results and SMAP reference data according to an embodiment of the present invention; Figure 6 It is a schematic diagram comparing the GNSS-R inversion result described in an embodiment of the present invention with a single-day two-dimensional image of SMAP parameter data. DETAILED DESCRIPTION

[0014] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0015] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and the like are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", and the like may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0016] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0017] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0018] like Figures 1 to 3 As shown, a multi-parameter GNSS-R soil moisture inversion method based on a random forest model includes the following steps: a) Quality control of spaceborne GNSS-R data Quality control adopts two methods: quality label and empirical criteria. First, the quality label data is converted into binary, and the data quality is determined according to the bit value. The label bits used mainly include black body load delay Doppler pattern mode, radio frequency interference, test mode, etc.; secondly, the data is screened again according to empirical criteria, and the variables and thresholds used mainly include: 1) the incident angle is greater than 65°, 2) the signal-to-noise ratio is less than 0, and 3) the antenna gain is less than 0. The data that meets the above conditions is screened out, and the main parameters are shown in Table 1.

[0019] Table 1 Main parameters of onboard GNSS-R data .

[0020] b) Extraction of spaceborne GNSS-R data observations and calculation of surface reflectivity The latitude and longitude coordinates of the satellite-borne GNSS-R mirror point and other observations used to calculate the surface reflectance are extracted. The surface reflectivity is calculated based on the bistatic radar equation. Assuming that coherent scattering mainly occurs on land, the coherent component is expressed in the following form: (1) In the formula, is the coherent scattering component power; the subscript r represents the right-hand circularly polarized antenna of the GNSS transmitter, and l represents the left-hand circularly polarized antenna of the GNSS-R receiver; is the signal wavelength; is the peak power of the transmitted signal; , are the gains of the transmitting antenna and the receiving antenna respectively; , are the distances between the transmitter, receiver and the mirror reflection point respectively; Represents the surface reflectivity.

[0021] The power expression of the incoherent scattered component is: (2) In the formula, is the incoherent scattering component power; Bistatic Radar Cross Section (BRCS), in m 2 .

[0022] make = , combine formula (1) and formula (2) to get the surface reflectivity writing Function, such as formula (3): (3) c) Accurately remove the spaceborne GNSS-R observation data affected by open water The seasonal products of the Global Surface Water Explorer (GSWE) dataset are used to mask the water areas, and the open water areas are accurately removed by coordinate matching based on the mask data. Figure 2 It is a seasonal product of GSW, which is reclassified and converted into a binary matrix form. 0 represents a waterless state, and 1 to 12 are marked as 1, representing a water state. In the case of missing data, a null value is assigned. According to the corresponding longitude and latitude information, the actual geographical location, area size and water state represented by each matrix element can be obtained, thereby obtaining the water body data used to mask the GNSS-R sampling points. The longitude and latitude of all GNSS-R sampling points in the study area are extracted and compared with the obtained geographical coordinates of the water body data. If the mirror point is located in the area covered by the open water in the mask data, it will be eliminated.

[0023] d) Gridding the GNSS-R surface reflectivity data like Figure 3 As shown in the figure, GNSS-R data are pseudo-random sampling points. In order to spatially match them with SMAP (Soil Moisture Active Passive (SMAP) satellite data) to construct a data set, the GNSS-R surface reflectivity is gridded using the EASE-Grid of the SMAP data, and the surface reflectivity of the sampling points falling within each grid is averaged to represent the surface reflectivity of the grid. Specifically, first, according to step c), the GNSS-R data falling in the open water area is removed; then the surface reflectivity of the GNSS-R data not falling in the open water area is calculated; finally, the GNSS-R surface reflectivity is gridded using the EASE-Grid of the SMAP data.

[0024] e) SMAP data preprocessing SMAP daily data contains soil moisture estimates, quality labels, vegetation optical thickness, land cover type, and grid-related coordinates for ascending and descending regions. The ascending and descending orbits are averaged to represent the results for one day. Similarly, in order to obtain global coverage data, the data for three consecutive days are averaged, ignoring null values. SMAP grid coordinates (a way of dividing the surface of the earth in SMAP data, which divides the surface into grids regularly, each grid represents an area with longitude and latitude coordinates), vegetation optical thickness, surface roughness, land cover type, and annual accumulation days are used as auxiliary data for the inversion model. SMAP soil moisture is used as the true value to verify the GNSS-R soil moisture inversion results. In addition, quality labels are used to filter SMAP data.

[0025] f) Inversion model construction All daily observation data are randomly divided into two groups, forming a training set and a test set, which account for 70% and 30% of the total data respectively. The 10-fold cross-validation method is used to tune the hyperparameters of the random forest model, which is the inversion model.

[0026] g) Soil moisture inversion and accuracy assessment using spaceborne GNSS-R The input features are substituted into the model to obtain the spaceborne GNSS-R soil moisture inversion results, and the quality of the results is evaluated from three perspectives: model accuracy, temporal variability, and spatial variability.

[0027] This method takes advantage of the performance of random forests, such as its ability to handle nonlinear relationships, good generalization performance, and strong robustness. After accurately eliminating the observation data affected by water bodies, it combines the multi-parameter auxiliary data of the Soil Moisture Active Passive (SMAP) satellite to construct an inversion model. This can significantly improve the spatiotemporal coverage, inversion accuracy, and reliability of soil moisture data, and provide strong data support for climate forecasting, agricultural production, hydrological modeling, and other fields.

[0028] Example 1 like Figure 4 , Figure 5 , Figure 6 Shown are the scatter plot of the test set data and the SMAP reference data, the random grid time series analysis graph, and the single-day two-dimensional image comparison graph.

[0029] in, Figure 4 The figure shows the scatter plot of the test set data and the SMAP reference data. The two have good consistency, with a correlation coefficient R of 0.928 and a root mean square error RMSE of 0.035 cm. 3 cm -3 .

[0030] in, Figure 5 The figure shows the time series analysis of the satellite GNSS-R inversion results and the SMAP reference data on a random grid. It can be seen that the correlation and error are maintained at a good level, the correlation coefficient R is not less than 0.92, the average value reaches 0.9396, and the root mean square error RMSE is not higher than 0.037cm 3 cm -3 , the average value reaches 0.0328cm 3 cm -3 .

[0031] in, Figure 6 The figure shows a comparison of the single-day two-dimensional image of the satellite-borne GNSS-R inversion result and the SMAP reference data. The correlation coefficient R can reach 0.959, and the root mean square error RMSE is 0.029cm 3 cm -3 .

[0032] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-parameter GNSS-R soil moisture inversion method based on a random forest model, characterized by: The following steps are involved: S1. Perform quality control on onboard GNSS-R data; S2, extracting the observed quantity of the satellite-borne GNSS-R data after quality control in step S1 and calculating the surface reflectivity to obtain the surface reflectivity, and proceeding to step S4; S3, eliminating the satellite-borne GNSS-R observation data affected by open waters, and proceeding to step S4; S4, gridding the surface reflectivity obtained in step S2 using the spaceborne GNSS-R observation data obtained in step S3 to obtain gridded data; S5, performing SMAP data preprocessing in combination with the gridded data in step S4; S6, constructing an inversion model; S7. Spaceborne GNSS-R soil moisture inversion and accuracy assessment.

2. The multi-parameter GNSS-R soil moisture inversion method based on the random forest model according to claim 1 is characterized in that: In step S1, the quality control of the spaceborne GNSS-R data is performed, including: Quality control is done by both quality labeling and empirical guidelines; First, the quality label data is converted into binary, and the data quality is determined based on the bit value. The label bits used include blackbody load delay Doppler pattern mode, radio frequency interference, and test mode. Secondly, the data was screened again according to empirical criteria, and the variables and threshold conditions used included: The incident angle is greater than 65°, the signal-to-noise ratio is less than 0, and the antenna gain is less than 0; Filter out the data that meets the conditions.

3. The multi-parameter GNSS-R soil moisture inversion method based on the random forest model according to claim 1 is characterized in that: In step S2, the observation quantity of the satellite-borne GNSS-R data after quality control in step S1 is extracted and the surface reflectivity is calculated to obtain the surface reflectivity, including: The latitude and longitude coordinates of the mirror point of the satellite-borne GNSS-R data and other observations used to calculate the surface reflectance are extracted. The surface reflectivity is calculated based on the bistatic radar equation. Assuming that coherent scattering occurs on land, the coherent component is expressed in the following form: (1) In the formula, is the coherent scattering component power; r represents the right-hand circularly polarized antenna of the GNSS transmitter, and l represents the left-hand circularly polarized antenna of the GNSS-R receiver; is the signal wavelength; is the peak power of the transmitted signal; , are the gains of the transmitting antenna and the receiving antenna respectively; , are the distances between the transmitter, receiver and the mirror reflection point respectively; represents the surface reflectivity; The power expression of the incoherent scattered component is: (2) In the formula, is the incoherent scattering component power; Bistatic Radar Cross Section (BRCS), in m 2 ; make = , combine formula (1) and formula (2) to get the surface reflectivity writing Function, such as formula (3): (3)。 4. The multi-parameter GNSS-R soil moisture inversion method based on random forest model according to claim 1, characterized in that: In step S3, the satellite-borne GNSS-R observation data affected by open waters are removed, including: The seasonal product of the global water dataset is used to mask the water area to obtain mask data, and the open water area is removed by coordinate matching based on the mask data; 0 represents the anhydrous state, and 1 to 12 are marked with 1 representing the water state; In case of missing data, a null value is assigned; Based on the longitude and latitude information, the geographical location, area size and water state represented by each matrix element can be obtained, thereby obtaining the water body data used to mask the GNSS-R sampling points; the longitude and latitude of all GNSS-R sampling points in the study area are extracted and compared with the obtained geographical coordinates of the water body data. If the mirror point is located in the area covered by the open water in the mask data, it will be eliminated.

5. The multi-parameter GNSS-R soil moisture inversion method based on random forest model according to claim 1, characterized in that: In step S4, the surface reflectivity obtained in step S2 is gridded using the spaceborne GNSS-R observation data obtained in step S3 to obtain gridded data, including: According to step S3, the GNSS-R data falling in the open water area is eliminated; and the surface reflectivity of the GNSS-R data not falling in the open water area is calculated, and then the surface reflectivity is gridded using the EASE-Grid of the SMAP data, and the surface reflectivity of the sampling points falling in each grid is averaged to represent the surface reflectivity of this grid.

6. The multi-parameter GNSS-R soil moisture inversion method based on random forest model according to claim 1, characterized in that: In step S5, SMAP data preprocessing is performed in combination with the gridded data in step S4, including: Daily SMAP data include soil moisture estimates, quality labels, vegetation optical thickness, land cover types, and grid-related coordinates of the ascending and descending regions. The grid-related coordinates of the ascending and descending regions are averaged to represent the results of one day. To obtain global coverage, data from three consecutive days were averaged, ignoring null values; SMAP grid coordinates, vegetation optical thickness, surface roughness, land cover type, and annual accumulation day are used as auxiliary data for the inversion model; The soil moisture of SMAP is used as the true value of the GNSS-R soil moisture inversion result in step S7; SMAP data were filtered using quality labels.

7. The multi-parameter GNSS-R soil moisture inversion method based on random forest model according to claim 1, characterized in that: In step S6, an inversion model is constructed, including: All daily observation data were randomly divided into two groups, forming a training set and a test set, which accounted for 70% and 30% of the total data, respectively. Ten-fold cross-validation was used to tune the hyperparameters of the random forest model.

8. The multi-parameter GNSS-R soil moisture inversion method based on random forest model according to claim 1, characterized in that: In step S7, the spaceborne GNSS-R soil moisture inversion and accuracy assessment includes: The input features are substituted into the inversion model to obtain the spaceborne GNSS-R soil moisture inversion results, and the quality of the results is evaluated from three perspectives: model accuracy, temporal variability, and spatial variability.

Images