Construction method, inversion method and device of intelligent constellation analysis system

Abstract

The invention relates to an intelligent constellation analysis system construction method, inversion method and device, and the method comprises the steps: carrying out the cross correction and time-space seamless filling of brightness temperature data of a target region, and obtaining a time-space consistent reference brightness temperature data set; extracting brightness temperature data corresponding to each sub-region from the reference brightness temperature data set, and screening out an optimal brightness temperature combination input set suitable for each sub-region; inputting the optimal brightness temperature combination input set of each sub-region into a plurality of snow water equivalent inversion algorithms, and determining a precision index of each snow water equivalent inversion algorithm; and sorting all the snow water equivalent inversion algorithms corresponding to each sub-region according to the precision index to form an optimal snow water equivalent inversion algorithm recommendation sequence of each sub-region, and constructing an intelligent constellation analysis system according to the optimal snow water equivalent inversion algorithm recommendation sequence of each sub-region. Therefore, a series of problems from data consistency to algorithm applicability are solved, and the precision of snow water equivalent inversion is remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of remote sensing technology, and in particular to a method for constructing, inverting, and assembling a smart constellation analysis system. Background Technology

[0002] Snow water equivalent (SWE) is a key parameter for measuring the liquid water content in snow cover, and it plays an irreplaceable role in hydrological cycle research, climate change research, water resource management, and snow disaster early warning. Passive microwave remote sensing, due to its ability to penetrate snow layers and its capacity for large-scale, high-frequency observations, has become a core technology for monitoring SWE on a global scale.

[0003] However, existing SWE inversion technology systems suffer from a series of inherent defects, severely limiting their accuracy and application value. Firstly, at the data foundation level, relying on observation data from a single satellite sensor (such as SSM / I or AMSR2) results in incomplete spatial coverage (orbital gaps), low temporal resolution, and the inability to construct long-term consistent datasets. Although the industry has proposed the concept of a "virtual constellation" to compensate by combining data from multiple satellites, the differences in system parameters such as center frequency, incident angle, and bandwidth between different sensors (such as SMMR, SSMI, AMSR2, and MWRI) lead to significant systematic biases (i.e., heterogeneity) in the observed brightness temperature data. Without effective unified correction, directly using these heterogeneous data sources will cause abrupt changes in the inversion results, making it impossible to form truly reliable long-term series products.

[0004] Secondly, regarding the inversion method, the snow characteristics (snow grain size, density, stratification) and underlying surface environments (forest cover, topography) vary greatly across different snow-covered regions globally (such as the Tibetan Plateau and the Arctic tundra). This means that no single "universal" SWE inversion algorithm can perform optimally in all regions. Existing technologies lack systematic analysis of the sensitivity of different bands and sensor combinations in different regions, and even more so, they lack an intelligent mechanism that can dynamically recommend the most suitable inversion algorithm based on regional characteristics. Users typically rely on experience or a single global algorithm for inversion, and the accuracy of the results in specific regions is often difficult to guarantee.

[0005] Therefore, there is an urgent need in this field for a technical solution that can fundamentally solve the problem of consistency of multi-source data and intelligently recommend the optimal inversion algorithm based on regional characteristics, so as to support the high-precision, long-time-series, and business-oriented SWE monitoring needs. Summary of the Invention

[0006] (a) Technical problems to be solved

[0007] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a method for constructing a smart constellation analysis system, an inversion method and a device, which solves the technical problem of low accuracy in the prior art.

[0008] (II) Technical Solution

[0009] To achieve the above objectives, the main technical solutions adopted by the present invention include:

[0010] In a first aspect, embodiments of the present invention provide a method for constructing a smart constellation analysis system, comprising: acquiring brightness temperature data from multiple passive microwave remote sensing sensors in a target area, and performing cross-correction and spatiotemporal seamless filling on the brightness temperature data of the target area to obtain a spatiotemporally consistent benchmark brightness temperature dataset; the target area includes multiple sub-regions; in response to the boundary information of each sub-region in the multiple sub-regions, extracting the brightness temperature data corresponding to each sub-region from the benchmark brightness temperature dataset, and selecting the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region; inputting the optimal brightness temperature combination input set of each sub-region into multiple snow water equivalent inversion algorithms, and determining the accuracy index of each snow water equivalent inversion algorithm based on the inversion results of multiple snow water equivalent inversion algorithms and the station observation data in the corresponding sub-regions; sorting all snow water equivalent inversion algorithms corresponding to each sub-region according to the accuracy index to form a recommended sequence of optimal snow water equivalent inversion algorithms for each sub-region; and constructing a smart constellation analysis system based on the recommended sequence of optimal snow water equivalent inversion algorithms for each sub-region.

[0011] In one possible embodiment, the plurality of passive microwave remote sensing sensors include at least two of SMMR, SSMI, SSMIS, AMSR2, and MWRI sensors.

[0012] In one possible embodiment, the linear regression model established during the cross-correction process is as follows:

[0013] ;

[0014] In the formula, Tb_cal represents the corrected brightness temperature data; slpoe represents the slope coefficient; Tb_ori represents the original brightness temperature data; and intercept represents the intercept coefficient.

[0015] In one possible embodiment, the spatiotemporal seamless filling process includes: for any spatiotemporal vacant pixel in the target area, the brightness temperature data of other satellites passing over within the same predetermined time window are preferentially used for filling; for vacant pixels that still exist after being filled by the brightness temperature data of other satellites, the brightness temperature data of the still existing vacant pixels are simulated and supplemented using a radiative transfer model that couples the dense medium radiative transfer model and the Qp soil emissivity model.

[0016] In one possible embodiment, the optimal brightness temperature combination input set suitable for each sub-region is selected from the brightness temperature data corresponding to each sub-region, including: based on the dense medium radiative transfer model, sensitivity analysis is performed on different bands combined with passive microwave remote sensing sensors to select the optimal brightness temperature combination input set suitable for the corresponding sub-region.

[0017] In one possible embodiment, the various snow water equivalent inversion algorithms include at least two of the following: Chang algorithm, Foster algorithm, Kelly algorithm, WESTDC algorithm, FY3 algorithm, snow radiation model combined with machine learning algorithm, and snow process model combined with machine learning algorithm.

[0018] In one possible embodiment, the accuracy metrics include at least root mean square error and bias.

[0019] In a second aspect, embodiments of the present invention provide a method for inverting regional snowmelt equivalent, comprising: acquiring boundary information of a sub-region to be predicted; searching for a target optimal snowmelt equivalent inversion algorithm recommendation sequence corresponding to the sub-region to be predicted from all optimal snowmelt equivalent inversion algorithm recommendation sequences stored in a smart constellation analysis system; and determining a target optimal snowmelt equivalent inversion algorithm corresponding to the sub-region to be predicted based on the target optimal snowmelt equivalent inversion algorithm recommendation sequence; wherein, the smart constellation analysis system is obtained by the construction method of the smart constellation analysis system as described in any of the first aspects; and calculating the snowmelt equivalent of the sub-region to be predicted based on the target optimal snowmelt equivalent inversion algorithm.

[0020] Thirdly, embodiments of the present invention provide a construction apparatus for a smart constellation analysis system, comprising:

[0021] The first acquisition module is used to acquire brightness temperature data of the target area from multiple passive microwave remote sensing sensors, and to perform cross-correction and spatiotemporal seamless filling on the brightness temperature data of the target area to obtain a spatiotemporally consistent benchmark brightness temperature dataset; the target area includes multiple sub-regions.

[0022] The extraction and filtering module is used to extract the brightness temperature data corresponding to each sub-region from the benchmark brightness temperature dataset in response to the boundary information of each sub-region in multiple sub-regions, and to filter the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region.

[0023] The input determination module is used to input the optimal brightness temperature combination input set of each sub-region into multiple snow water equivalent inversion algorithms, and to determine the accuracy index of each snow water equivalent inversion algorithm based on the inversion results of multiple snow water equivalent inversion algorithms and the station observation data in the corresponding sub-region.

[0024] The sorting module is used to sort all the snow water equivalent inversion algorithms corresponding to each sub-region according to the accuracy index, so as to form the recommended sequence of the optimal snow water equivalent inversion algorithm for each sub-region;

[0025] The module is used to build a smart constellation analysis system based on the recommended sequence of the optimal snow water equivalent inversion algorithm for each sub-region.

[0026] Fourthly, embodiments of the present invention provide a regional snowmelt equivalent inversion device, comprising:

[0027] The second acquisition module is used to acquire the boundary information of the sub-region to be predicted.

[0028] The lookup and determination module is used to find the target optimal snow water equivalent inversion algorithm recommendation sequence corresponding to the sub-region to be predicted from the optimal snow water equivalent inversion algorithm recommendation sequences of all sub-regions stored in the intelligent constellation analysis system, and to determine the target optimal snow water equivalent inversion algorithm corresponding to the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm recommendation sequence; wherein, the intelligent constellation analysis system is obtained by the construction method of the intelligent constellation analysis system as described in any of the first aspects;

[0029] The calculation module is used to calculate the snow water equivalent of the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm.

[0030] (III) Beneficial Effects

[0031] The beneficial effects of this invention are:

[0032] This invention proposes a method, inversion method, and apparatus for constructing a smart constellation analysis system. It acquires brightness temperature data from multiple passive microwave remote sensing sensors for a target region, performs cross-correction and spatiotemporal seamless filling on the brightness temperature data to obtain a spatiotemporally consistent baseline brightness temperature dataset. Responding to the boundary information of each sub-region within multiple sub-regions, it extracts the brightness temperature data corresponding to each sub-region from the baseline brightness temperature dataset, and filters the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region. Finally, it sets the optimal brightness temperature combination for each sub-region... The input set is fed into multiple snow water equivalent inversion algorithms. Based on the inversion results of the multiple snow water equivalent inversion algorithms and the station observation data in the corresponding sub-regions, the accuracy index of each snow water equivalent inversion algorithm is determined. All snow water equivalent inversion algorithms corresponding to each sub-region are sorted according to the accuracy index to form the optimal snow water equivalent inversion algorithm recommendation sequence for each sub-region. Based on the optimal snow water equivalent inversion algorithm recommendation sequence for each sub-region, a smart constellation analysis system is constructed, thereby significantly improving the accuracy, reliability and automation of snow water equivalent inversion.

[0033] To make the above-mentioned objectives, features and advantages to be achieved by the embodiments of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0034] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 A flowchart illustrating a method for constructing a smart constellation analysis system according to an embodiment of this application is shown;

[0036] Figure 2 This paper shows a structural block diagram of a radiative transfer model provided in an embodiment of the present application;

[0037] Figure 3 A flowchart of a regional snowmelt equivalent inversion method provided in an embodiment of this application is shown;

[0038] Figure 4 This paper shows a structural block diagram of a smart constellation analysis system construction device 400 provided in an embodiment of this application;

[0039] Figure 5 A structural block diagram of a regional snowmelt equivalent inversion device 500 provided in an embodiment of this application is shown. Detailed Implementation

[0040] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] This invention proposes a method, inversion method, and apparatus for constructing a smart constellation analysis system. It acquires brightness temperature data from multiple passive microwave remote sensing sensors for a target region, performs cross-correction and spatiotemporal seamless filling on the brightness temperature data to obtain a spatiotemporally consistent baseline brightness temperature dataset. Responding to the boundary information of each sub-region within multiple sub-regions, it extracts the brightness temperature data corresponding to each sub-region from the baseline brightness temperature dataset, and filters the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region. Finally, it sets the optimal brightness temperature combination for each sub-region... The input set is fed into multiple snow water equivalent inversion algorithms. Based on the inversion results of the multiple snow water equivalent inversion algorithms and the station observation data in the corresponding sub-regions, the accuracy index of each snow water equivalent inversion algorithm is determined. All snow water equivalent inversion algorithms corresponding to each sub-region are sorted according to the accuracy index to form the optimal snow water equivalent inversion algorithm recommendation sequence for each sub-region. Based on the optimal snow water equivalent inversion algorithm recommendation sequence for each sub-region, a smart constellation analysis system is constructed, thereby significantly improving the accuracy, reliability and automation of snow water equivalent inversion.

[0042] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0043] First Embodiment

[0044] Please see Figure 1 , Figure 1 A flowchart illustrating a method for constructing a smart constellation analysis system according to an embodiment of this application is shown. It should be understood that this construction method can be executed by a device for constructing the smart constellation analysis system, and the specific device can be configured according to actual needs; this embodiment is not limited thereto. For example, the device can be a computer or a server, etc. Specifically, the construction method includes:

[0045] Step S110: Acquire brightness temperature data of the target area from multiple passive microwave remote sensing sensors, and perform cross-correction and spatiotemporal seamless filling on the brightness temperature data of the target area to obtain a spatiotemporally consistent benchmark brightness temperature dataset. The target area includes multiple sub-regions.

[0046] It should be understood that the specific area of ​​the target area can be set according to actual needs, and the embodiments of this application are not limited thereto.

[0047] It should also be understood that the specific sensors of the multiple passive microwave remote sensing sensors can be set according to actual needs, and the embodiments of this application are not limited thereto.

[0048] Optionally, the plurality of passive microwave remote sensing sensors include at least two of the following: Scanning Multichannel Microwave Radiometer (SMMR), Special Sensor Microwave Imager (SSMI), Special Sensor Microwave Imager / Sounder (SSMIS), Advanced Microwave Scanning Radiometer 2 (AMSR2), and Microwave Radiation Imager (MWRI).

[0049] It should also be noted that the brightness temperature data from the aforementioned multiple passive microwave remote sensing sensors can be within a specified time period, and the specific time period can be set according to actual needs; the embodiments of this application are not limited to this. For example, the specified time period could be from 1978 to 2024.

[0050] It should also be understood that the specific process of cross-correcting the brightness temperature data of the target area can be set according to actual needs, as long as it can eliminate the systematic deviation between different sensors. The embodiments of this application are not limited to this.

[0051] Optionally, this cross-correction employs a pixel-by-pixel linear regression method, using brightness temperature data from one satellite during the time overlap period as a reference to correct brightness temperature data from the same band of another satellite, establishing the following linear regression model:

[0052] ;

[0053] In the formula, Tb_cal represents the corrected brightness temperature data; slpoe represents the slope coefficient; Tb_ori represents the original brightness temperature data; and intercept represents the intercept coefficient.

[0054] For example, firstly, data preparation and pairing can be performed, which includes: collecting all available passive microwave satellite data (such as SMMR, SSMI, SSMIS, AMSR2, and MWRI), and identifying sensor pairs whose data have overlapping observation periods in time, such as FY-3D / MWRI and GCOM-W / AMSR2, which operated simultaneously between 2018 and 2020. These overlapping periods are the basis for corrections.

[0055] Subsequently, pixel-by-pixel regression analysis can be performed, which includes: the system processing each spatial pixel (e.g., a 0.25° × 0.25° grid) within the target area individually; and, for a given pixel and a specific frequency / polarization (e.g., 36.5 GHz V polarization), the system extracts all synchronous or quasi-synchronous observations of that pixel from two satellites (e.g., Sensor_A and Sensor_B) during the temporal overlap period; and, using the brightness temperature of a stable satellite considered a "benchmark" (e.g., AMSR2) as the dependent variable and the brightness temperature of the satellite to be corrected (e.g., MWRI) as the independent variable, performing linear regression analysis.

[0056] Subsequently, a correction coefficient lookup table can be generated: the linear regression model above generates unique slope and intercept coefficients for each pixel, each sensor pair, and each band, and all these coefficients are stored in a "correction coefficient lookup table" covering the target area.

[0057] Finally, in subsequent processing, for any satellite's raw brightness temperature data, the system will retrieve the corresponding correction coefficient from the "correction coefficient lookup table" based on its sensor type, band, and pixel location, and calculate the corrected brightness temperature value that is consistent with the reference sensor standard using the formula of the linear regression model.

[0058] Therefore, with the help of the above technical solution, all historical data of satellites are unified under the same brightness temperature "metric", which solves the "heterogeneity" problem and enables direct numerical comparison and fusion of data from different sources.

[0059] It should also be understood that the specific process of seamlessly filling the missing pixels in the target area can be set according to actual needs, as long as it can eliminate the systematic deviation between different sensors. The embodiments of this application are not limited to this.

[0060] Optionally, for any spatiotemporally missing pixel within the target area, brightness temperature data from other satellites passing over within the same predetermined time window are preferentially used to fill the gap. For missing pixels that still exist after being filled with brightness temperature data from other satellites, a radiative transfer model coupled with a dense medium radiative transfer model (DMRT) and a Qp soil emissivity model is used to simulate and supplement the brightness temperature data of the remaining missing pixels. Here, the missing pixel is a composite concept, describing a location lacking valid observation data, and can be described from three dimensions: spatial dimension, temporal dimension, and data state. The spatial dimension refers to a specific, smallest unit with a clearly defined geographic coordinate range, i.e., a pixel or grid; the temporal dimension refers to a specific, discrete observation time point or time period; and the data state refers to the fact that at the aforementioned specific spatial location and specific time point, no satellite provides valid, usable passive microwave brightness temperature observation values.

[0061] For example, considering the gaps between the orbits of individual satellites, and the situation where certain areas on Earth lack observation data at a specific time, when a satellite (such as satellite A) has an observation gap over a target area at a certain time, the system will automatically search for data from other satellites (such as satellite B) passing over at the same or similar time. Since these data have been cross-corrected to eliminate system differences, the spatial gap of satellite A can be directly filled with data from other satellites. Considering that even after using a multi-satellite combination, some pixels may still not have any real observation values ​​from any satellite at a certain point in time due to extreme weather, lack of coverage by all satellites, or other reasons, the system will activate a radiative transfer model that couples a dense medium radiative transfer model and a Qp model (simulating soil). This radiative transfer model will combine the reanalysis data at that time (such as ERA5, which provides atmospheric and surface conditions) and background knowledge to simulate and calculate the brightness temperature value of the missing pixel at that time.

[0062] It should also be understood that the specific model structure of the radiative transfer model can be set according to actual needs, as long as it is coupled with the dense medium radiative transfer model and the Qp model. The embodiments of this application are not limited to this.

[0063] Optionally, such as Figure 2 As shown, the radiative transfer model includes:

[0064] The snow layer DMRT enhanced calculation module 210 integrates a dense medium radiative transfer model, which is used to calculate the volume scattering, absorption and self-emission brightness temperature contribution of the snow layer by dividing the snow layer into multiple sublayers and solving the radiative transfer equation.

[0065] The soil Qp model optimization module 220 integrates a Qp soil emissivity model, which is used to calculate the emission and scattering brightness temperature contribution of rough soil surfaces by calculating the frequency-adaptive Qp factor and soil dielectric properties.

[0066] The snow-soil interaction scattering coordination module 230 is connected to the snow layer DMRT enhancement calculation module 210 and the soil Qp model optimization module 220, respectively. It is used to couple the output brightness temperature contribution of the two through a dynamic weight allocation mechanism and calculate the interaction scattering contribution between the snow layer and the soil interface.

[0067] Specifically, the snow layer DMRT enhancement calculation module 210 can be used to: vertically divide the snow layer into multiple sublayers, and calculate the snow particle scattering and absorption characteristics of each sublayer based on snow water equivalent, snow particle size distribution, and snow layer temperature profile; based on the scattering and absorption characteristics of each sublayer, calculate the multiple scattering and transmission process of electromagnetic waves in the snow layer by solving the one-dimensional vector radiative transfer equation, and the snow particle scattering and absorption characteristics of each sublayer can be obtained by calculation using the improved Mie scattering theory, wherein the dielectric constant of ice particles includes viscous attenuation correction; integrate the radiative transfer solutions of all sublayers to obtain the comprehensive contribution of the entire snow layer's self-emission and internal multiple scattering, denoted as Tb_DMRT;

[0068] The soil Qp model optimization module 220 can be specifically used to: obtain the soil moisture, root mean square height s, correlation length l, and soil texture type of the target pixel; based on the root mean square height s and the incident angle... Through formula Calculate the frequency-adaptive Qp factor, where the incident angle is... When observing the Earth's surface using a satellite sensor, the angle between the propagation direction of the electromagnetic beam and the surface normal is defined as follows: f is the center frequency of the operating channel of the passive microwave satellite sensor, and c is the speed of light. Based on soil moisture, soil texture type, and Qp factor, the polarization emissivity of the rough soil surface is calculated using the Qp model. Combining the surface physical temperature, the polarization emissivity is converted into brightness temperature to obtain the combined contribution of soil volume emission and surface scattering, denoted as Tb_Qp.

[0069] The snow-soil interaction scattering coordination module 230 can be specifically used to: calculate the snow-soil interaction scattering contribution Tb_interaction based on snow transmittance t and snow reflectance r using the formula Tb_interaction=r×Tb_Qp×t+Rs, where snow transmittance t refers to the proportion of electromagnetic wave energy incident on the snow surface that can penetrate the entire snow layer and reach the soil-snow interface, snow reflectance r refers to the proportion of electromagnetic wave energy incident on the snow surface that is backscattered back into the atmosphere by the entire snow layer (including the upper surface and internal volume of the snow layer), and Rs is the reflection contribution of downward radiation within the snow layer; and determine the weighting coefficient based on the microwave frequency. , and And using the formula Tb_total= ×Tb_DMRT+ ×Tb_Qp+ ×Tb_interaction calculates the final brightness temperature, where the weighting coefficients satisfy... + + =1 energy conservation constraint.

[0070] Step S120: In response to the boundary information of each sub-region in multiple sub-regions, extract the brightness temperature data corresponding to each sub-region from the benchmark brightness temperature dataset, and select the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region.

[0071] It should be understood that the specific process of selecting the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region can be set according to actual needs, and the embodiments of this application are not limited thereto.

[0072] Optionally, based on the dense medium radiative transfer model, sensitivity analysis is performed on different bands and passive microwave remote sensing sensor combinations to select the optimal brightness temperature combination input set suitable for the corresponding sub-region. The dense medium radiative transfer model can be an existing model; the sensitivity analysis determines which frequency, polarization, and sensor combination is most sensitive and effective for the snow cover characteristics of the current sub-region.

[0073] For example, based on the Dense Medium Radiative Transfer Model (DMRT), the system simulates the response of the current sub-region to various brightness temperature combinations under different snow conditions (different snow depths, snow densities, and snow grain sizes). By analyzing these simulation data, the system quantifies the sensitivity relationship between each brightness temperature combination and snow water equivalent (e.g., calculating correlation coefficients and sensitivity coefficients). Finally, it selects one or more brightness temperature combinations that are most sensitive to changes in snow water equivalent and have the highest thresholds approaching saturation as the optimal brightness temperature combination input set for the current sub-region. For example, the analysis results may show that in the current sub-region with deep snow, the low-frequency 10.65V and 36.5V combination has stronger penetrating power than the high-frequency combination, and is therefore selected as one of the optimal combinations.

[0074] Step S130: Input the optimal brightness temperature combination input set of each sub-region into multiple snow water equivalent inversion algorithms, and determine the accuracy index of each snow water equivalent inversion algorithm based on the inversion results of multiple snow water equivalent inversion algorithms and the station observation data in the corresponding sub-region.

[0075] It should be understood that the algorithms included in various snow water equivalent inversion algorithms can be set according to actual needs, and the embodiments of this application are not limited thereto.

[0076] Optionally, the various snow water equivalent inversion algorithms include at least two of the following: Chang algorithm, Foster algorithm, Kelly algorithm, WESTDC algorithm, FY3 algorithm, snow radiation model combined with machine learning algorithm, and snow process model combined with machine learning algorithm.

[0077] Furthermore, the computational expression for Chang's algorithm is:

[0078] ;

[0079] Where SD represents the inverted snow depth; The brightness temperature is 19 or 18.7 GHz H-polarized; The brightness temperature is 37 or 36.5 GHz H-polarization;

[0080] The computational expression for the Foster algorithm is as follows:

[0081] ;

[0082] in, The brightness temperature is 19 or 18.7 GHz H-polarized; 37 or 36.5 GHz H-polarization brightness temperature; ff is forest cover;

[0083] The Kelly algorithm's calculation expression is as follows:

[0084] ;

[0085] ;

[0086] ;

[0087] Where SD represents the inverted snow depth; SD open Snow depth retrieved for non-forest cover areas; SD forest pol represents the snow depth inverted from the forest cover area; ff represents the forest cover; pol represents the snow depth inverted from the forest cover area. 19 The brightness temperature difference between H-polarization and V-polarization is 19 or 18.7 GHz; pol 37 The brightness temperature difference between H-polarization and V-polarization at 37 or 36.5 GHz; fd is the forest density;

[0088] The calculation expression for the WESTDC algorithm is as follows:

[0089] ;

[0090] Where SD represents the inverted snow depth; The brightness temperature is 19 or 18.7 GHz H-polarized; ff represents the brightness temperature of the 37 or 36.5 GHz H polarization; ff represents the forest cover.

[0091] The calculation expression for the FY3 algorithm is as follows:

[0092] ;

[0093] Where SD represents the inverted snow depth; SD grass Snow depth in the grassland pixels; SD forest Snow depth for forest pixels; SD barren Snow depth for bare ground pixels; SD farm The snow in the farmland was as deep as that of Yuan; ff grass ff forest ff barren ff farm These represent the coverage of grassland, forest, bare land, and farmland within a pixel, respectively.

[0094] The calculation expression for this snow radiation model combined with machine learning algorithm is as follows:

[0095] ;

[0096] Where SD represents the retrieved snow depth; RF represents the random forest model; and TBD represents the TBD model. 18.7V&36.5V and TBD 10.65V&36.5VThe values ​​represent the brightness temperature differences between 18.7 GHz V-polarization and 36.5 GHz V-polarization, and between 10.65 GHz V-polarization and 36.5 GHz V-polarization, respectively; longitude is longitude; elevation is altitude; effGS is effective snow particle size, which is obtained by minimizing the difference between the brightness temperature difference between 18.7 GHz and 36.5 GHz simulated by the HUT model and the brightness temperature difference observed by actual satellites.

[0097] The computational expression for this snow accumulation process model combined with machine learning algorithm is as follows:

[0098] :

[0099] Where SD represents the retrieved snow depth; RF represents the random forest model; and TBD represents the TBD model. 18.7V&36.5V and TBD 10.65V&18.7V The brightness temperature differences are 18.7 GHz V-polarization and 36.5 GHz V-polarization, and 10.65 GHz V-polarization and 18.7 GHz V-polarization, respectively; longitude is longitude; latitude is latitude; elevation is altitude; fc is forest cover; SD SNTHERM The snow depth is simulated using the SNTHERM model; D avg It is the average snow grain size simulated by the SNTHERM model; D eff It is at 36.5 GHz, and was obtained by optimizing MEMLS model based on 10 layers of SNTHERM data; It is the average value of the multi-layer snow density simulated by the SNTHERM model.

[0100] In addition, the station observation data for this target area may include data from the Global Historical Climatology Network (GHCN) and the Snow Telemetry (SNOTEL) system.

[0101] In addition, this accuracy indicator includes at least the root mean square error and the deviation, and its calculation expression is as follows:

[0102] ;

[0103] ;

[0104] Where RMSE is the root mean square error; N is the total number of samples; Let be the equivalent of the snowmelt inverted i-th time; The measured snow water equivalent for the corresponding station; Bias represents the deviation.

[0105] Step S140: Sort all snow water equivalent inversion algorithms corresponding to each sub-region according to the accuracy index to form the recommended sequence of optimal snow water equivalent inversion algorithms for each sub-region.

[0106] It should be understood that the specific process of sorting all snow water equivalent inversion algorithms corresponding to each sub-region according to the accuracy index can also be set according to actual needs, and the embodiments of this application are not limited thereto.

[0107] For example, prioritize sorting by root mean square error (RMSE) from smallest to largest. For algorithms with the same RMSE, sort by absolute value of bias from smallest to largest.

[0108] For example, the values ​​can be sorted from smallest to largest by a weighted composite index of root mean square error (RMSE) and bias, with the weight of RMSE being greater than that of bias.

[0109] Step S150: Construct a smart constellation analysis system based on the recommended sequence of the optimal snow water equivalent inversion algorithm for each sub-region.

[0110] It should be noted that this intelligent constellation analysis system incorporates the selection and combination of different satellite sensors, as well as the combination of model simulation and other methods, to provide seamless satellite data in any region, while also providing recommended sequences for snow water equivalent inversion algorithms.

[0111] In summary, by utilizing the above technical solutions, the construction method of the intelligent constellation analysis system proposed in this application has the following technical advantages:

[0112] A highly consistent multi-source data foundation was established: by introducing cross-correction and spatiotemporal seamless filling techniques, the systematic bias between different passive microwave sensors was successfully eliminated, and the problems of incomplete spatial coverage and time series interruptions were solved, generating a set of benchmark brightness temperature datasets with wide coverage, long time series, and high consistency. This provides reliable and unified data input for subsequent high-precision inversion, fundamentally ensuring the quality of long-time series analysis products;

[0113] This application achieves regionally adaptive optimal input selection: by selecting the optimal brightness temperature combination input set for each sub-region, it overcomes the limitations of a single global frequency combination. This method can automatically identify the brightness temperature bands and polarization combinations most sensitive to the physical properties of snow cover in a specific region, thus providing the most informative and discriminative input features for subsequent inversion, significantly improving key performance aspects such as penetration in deep snow areas and sensitivity in shallow snow areas.

[0114] An objective and quantitative algorithm recommendation mechanism is provided: by running multiple mainstream inversion algorithms in parallel and using site observation data to verify and rank their accuracy, this application transforms algorithm selection from subjective experience-based judgment to an objective data-driven decision-making process. Ultimately, a quantitative optimal algorithm recommendation sequence is generated for each sub-region, enabling users to directly adopt the best-performing algorithm in that region for inversion, greatly ensuring the reliability and accuracy of the SWE inversion results.

[0115] A complete intelligent technology chain has been built: This application realizes one-stop output from raw multi-source heterogeneous data to the final customized inversion scheme, which greatly improves the automation and intelligence level of SWE inversion business and provides powerful and easy-to-use technical tools for applications in hydrology, climate and other fields.

[0116] It should be understood that the above-described method for constructing a smart constellation analysis system is merely exemplary, and those skilled in the art can make various modifications based on the above method, and the modified solutions also fall within the protection scope of this application.

[0117] Second Embodiment

[0118] Please see Figure 3 , Figure 3 A flowchart of a regional snowmelt equivalent inversion method provided in an embodiment of this application is shown. It should be understood that this regional snowmelt equivalent inversion method can be executed by a regional snowmelt equivalent inversion device, and the specific device can be configured according to actual needs; this embodiment is not limited thereto. For example, the regional snowmelt equivalent inversion device can be a computer, or a server, etc. Specifically, the regional snowmelt equivalent inversion method includes:

[0119] Step S310: Obtain the boundary information of the sub-region to be predicted;

[0120] Step S320: Find the target optimal snow water equivalent inversion algorithm recommendation sequence corresponding to the sub-region to be predicted from the optimal snow water equivalent inversion algorithm recommendation sequences of all sub-regions stored in the smart constellation analysis system, and determine the target optimal snow water equivalent inversion algorithm corresponding to the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm recommendation sequence; wherein, the smart constellation analysis system is obtained by the construction method of the smart constellation analysis system as described in the first embodiment;

[0121] Step S330: Calculate the snow water equivalent of the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm.

[0122] Third Embodiment

[0123] Please see Figure 4 , Figure 4A structural block diagram of a smart constellation analysis system construction apparatus 400 provided in an embodiment of this application is shown. It should be understood that the smart constellation analysis system construction apparatus 400 is capable of performing the various steps in the above method embodiments. The specific functions of the smart constellation analysis system construction apparatus 400 can be found in the description above; detailed descriptions are omitted here to avoid repetition. The smart constellation analysis system construction apparatus 400 includes at least one software functional module that can be stored in a memory or embedded in the operating system (OS) of the smart constellation analysis system construction apparatus 400 in the form of software or firmware. Specifically, the construction apparatus 400 includes:

[0124] The first acquisition module 410 is used to acquire brightness temperature data of the target area from multiple passive microwave remote sensing sensors, and to perform cross-correction and spatiotemporal seamless filling on the brightness temperature data of the target area to obtain a spatiotemporally consistent benchmark brightness temperature dataset; the target area includes multiple sub-regions.

[0125] The extraction and filtering module 420 is used to extract the brightness temperature data corresponding to each sub-region from the benchmark brightness temperature dataset in response to the boundary information of each sub-region in multiple sub-regions, and to filter the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region.

[0126] The input determination module 430 is used to input the optimal brightness temperature combination input set of each sub-region into multiple snow water equivalent inversion algorithms, and determine the accuracy index of each snow water equivalent inversion algorithm based on the inversion results of multiple snow water equivalent inversion algorithms and the station observation data in the corresponding sub-region.

[0127] The sorting module 440 is used to sort all the snow water equivalent inversion algorithms corresponding to each sub-region according to the accuracy index, so as to form the optimal snow water equivalent inversion algorithm recommendation sequence for each sub-region;

[0128] Module 450 is used to build a smart constellation analysis system based on the recommended sequence of the optimal snow water equivalent inversion algorithm for each sub-region.

[0129] Since the apparatus described in the above embodiments of the present invention is an apparatus used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and variations of the apparatus based on the methods described in the above embodiments of the present invention, and therefore will not be described again here. All apparatuses used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0130] Fourth embodiment

[0131] Please see Figure 5 , Figure 5 A structural block diagram of a regional snowmelt equivalent inversion device 500 provided in an embodiment of this application is shown. It should be understood that the regional snowmelt equivalent inversion device 500 is capable of performing the various steps in the above method embodiments. The specific functions of the regional snowmelt equivalent inversion device 500 can be found in the description above; detailed descriptions are omitted here to avoid repetition. The regional snowmelt equivalent inversion device 500 includes at least one software function module that can be stored in a memory or embedded in the operating system (OS) of the regional snowmelt equivalent inversion device 500 in the form of software or firmware. Specifically, the regional snowmelt equivalent inversion device 500 includes:

[0132] The second acquisition module 510 is used to acquire the boundary information of the sub-region to be predicted;

[0133] The lookup and determination module 520 is used to find the target optimal snow water equivalent inversion algorithm recommendation sequence corresponding to the sub-region to be predicted from the optimal snow water equivalent inversion algorithm recommendation sequences of all sub-regions stored in the smart constellation analysis system, and determine the target optimal snow water equivalent inversion algorithm corresponding to the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm recommendation sequence; wherein, the smart constellation analysis system is obtained by the construction method of the smart constellation analysis system as in the first embodiment;

[0134] The calculation module 530 is used to calculate the snow water equivalent of the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm.

[0135] Since the apparatus described in the above embodiments of the present invention is an apparatus used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and variations of the apparatus based on the methods described in the above embodiments of the present invention, and therefore will not be described again here. All apparatuses used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0136] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0137] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0138] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0139] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0140] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0141] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A method for constructing an intelligent constellation analysis system, characterized in that, include: Brightness temperature data from multiple passive microwave remote sensing sensors are acquired for the target region, and cross-correction and spatiotemporal seamless filling are performed on the brightness temperature data of the target region to obtain a spatiotemporally consistent benchmark brightness temperature dataset; the target region includes multiple sub-regions. In response to the boundary information of each of the multiple sub-regions, the brightness temperature data corresponding to each sub-region is extracted from the benchmark brightness temperature dataset, and the optimal brightness temperature combination input set suitable for each sub-region is selected from the brightness temperature data corresponding to each sub-region. The optimal brightness temperature combination input set of each sub-region is input into multiple snow water equivalent inversion algorithms, and the accuracy index of each snow water equivalent inversion algorithm is determined based on the inversion results of the multiple snow water equivalent inversion algorithms and the station observation data in the corresponding sub-region. All snow water equivalent inversion algorithms corresponding to each sub-region are sorted according to the accuracy index to form a recommended sequence of optimal snow water equivalent inversion algorithms for each sub-region; The intelligent constellation analysis system is constructed based on the recommended sequence of the optimal snow water equivalent inversion algorithm for each sub-region.

2. The construction method according to claim 1, characterized in that, The plurality of passive microwave remote sensing sensors include at least two of the following: SMMR, SSMI, SSMIS, AMSR2, and MWRI sensors.

3. The construction method according to claim 1 or 2, characterized in that, The linear regression model established during the cross-correction process is as follows: ； In the formula, Tb_cal represents the corrected brightness temperature data; slpoe represents the slope coefficient; Tb_ori represents the original brightness temperature data; and intercept represents the intercept coefficient.

4. The construction method according to claim 3, characterized in that, The process of seamless spatiotemporal filling includes: For any spatiotemporal missing pixel in the target area, brightness temperature data from other satellites passing over within the same predetermined time window are used to fill it in first. For the missing pixels that still exist after the brightness temperature data from the other satellites are filled in, the brightness temperature data of the missing pixels is simulated and supplemented using a radiative transfer model that couples the dense medium radiative transfer model and the Qp soil emissivity model.

5. The construction method according to claim 1, characterized in that, The step of selecting the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region includes: Based on the dense medium radiative transfer model, sensitivity analysis was performed on different combinations of passive microwave remote sensing sensors with different wavebands to select the optimal brightness temperature combination input set suitable for the corresponding sub-region.

6. The construction method according to claim 1, characterized in that, The various snow water equivalent inversion algorithms include at least two of the following: Chang algorithm, Foster algorithm, Kelly algorithm, WESTDC algorithm, FY3 algorithm, snow radiation model combined with machine learning algorithm, and snow process model combined with machine learning algorithm.

7. The construction method according to claim 1, characterized in that, The accuracy indicators include at least root mean square error and deviation.

8. A method for inverting regional snowmelt equivalent, characterized in that, include: Obtain the boundary information of the sub-region to be predicted; The system retrieves the target optimal snow water equivalent inversion algorithm recommendation sequence corresponding to the sub-region to be predicted from all the optimal snow water equivalent inversion algorithm recommendation sequences stored in the intelligent constellation analysis system, and determines the target optimal snow water equivalent inversion algorithm corresponding to the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm recommendation sequence; wherein, the intelligent constellation analysis system is obtained by the construction method of the intelligent constellation analysis system as described in any one of claims 1 to 7; Based on the target optimal snow water equivalent inversion algorithm, the snow water equivalent of the sub-region to be predicted is calculated.

9. A device for constructing a smart constellation analysis system, characterized in that, include: The first acquisition module is used to acquire brightness temperature data of the target area from multiple passive microwave remote sensing sensors, and to perform cross-correction and spatiotemporal seamless filling on the brightness temperature data of the target area to obtain a spatiotemporally consistent benchmark brightness temperature dataset; the target area includes multiple sub-regions. The extraction and filtering module is used to extract the brightness temperature data corresponding to each sub-region from the benchmark brightness temperature dataset in response to the boundary information of each sub-region in the plurality of sub-regions, and to filter the optimal brightness temperature combination input set suitable for each sub-region from the brightness temperature data corresponding to each sub-region; The input determination module is used to input the optimal brightness temperature combination input set of each sub-region into multiple snow water equivalent inversion algorithms, and determine the accuracy index of each snow water equivalent inversion algorithm based on the inversion results of the multiple snow water equivalent inversion algorithms and the station observation data in the corresponding sub-region. The sorting module is used to sort all the snow water equivalent inversion algorithms corresponding to each sub-region according to the accuracy index, so as to form the optimal snow water equivalent inversion algorithm recommendation sequence for each sub-region; The module is used to construct the smart constellation analysis system based on the recommended sequence of the optimal snow water equivalent inversion algorithm for each sub-region.

10. A regional snowmelt equivalent inversion device, characterized in that, include: The second acquisition module is used to acquire the boundary information of the sub-region to be predicted. The lookup and determination module is used to find the target optimal snow water equivalent inversion algorithm recommendation sequence corresponding to the sub-region to be predicted from the optimal snow water equivalent inversion algorithm recommendation sequences of all sub-regions stored in the smart constellation analysis system, and to determine the target optimal snow water equivalent inversion algorithm corresponding to the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm recommendation sequence; wherein, the smart constellation analysis system is obtained by the construction method of the smart constellation analysis system as described in any one of claims 1 to 7; The calculation module is used to calculate the snow water equivalent of the sub-region to be predicted based on the target optimal snow water equivalent inversion algorithm.

Images