Troposphere delay product generation method combining GNSS (Global Navigation Satellite System) and polar orbit remote sensing satellite

Abstract

The invention relates to the technical field of high-precision positioning, and discloses a GNSS and polar orbit remote sensing satellite combined troposphere delay product generation method, which comprises the following steps: calculating a difference value between a real-time GNSS true value point and a remote sensing satellite troposphere delay background field to generate a group of differential anchor points, and after reliability arbitration is carried out on the differential anchor points, generating a group of differential anchor points; a continuous troposphere difference correction field is generated through a topographic constraint interpolation algorithm, the weight basis of the interpolation algorithm is an effective path distance considering topographic obstacle punishment, and finally the correction field is superposed on a background field. According to the method, direct mathematical fitting of data of different sources in the prior art is replaced, a terrain constraint interpolation mode is utilized, so that spatial propagation of correction information can follow limitation of a real physical barrier, and a correction result which does not conform to reality is prevented from being generated in a complex terrain area.

Description

Technical Field

[0001] This invention relates to a method for generating tropospheric delay products by combining GNSS and polar-orbiting remote sensing satellites, belonging to the field of high-precision positioning technology. Background Technology

[0002] To achieve centimeter-level or even more precise positioning, it is essential to effectively correct for the delay caused by atmospheric refraction in the signal propagation path. Among these, the total tropospheric zenith delay is one of the main sources of error. To obtain spatiotemporally continuous tropospheric delay products, a common technique in the industry is to fuse the high-precision point values ​​retrieved in real time from sparsely distributed GNSS reference stations on the ground with the background field obtained from polar-orbiting remote sensing satellites, which has spatial continuity but a time delay of several hours, using mathematical algorithms such as Kriging interpolation or weighted averaging.

[0003] However, when this technology is applied to scenarios with higher requirements for positioning continuity and reliability, such as automated logistics by drones in hilly areas, its inherent limitations are exposed. The effectiveness of this method relies on a core assumption that the changes in the troposphere are a spatially continuous and slow, stable field. However, in areas with rapidly changing terrain, the distribution of tropospheric water vapor often exhibits spatial jumps due to the influence of local circulation. At the same time, under dynamic weather conditions, the temporal evolution of tropospheric delay is also nonlinear. This disconnect between theoretical assumptions and physical reality leads to insufficient accuracy of the tropospheric products generated in the above scenarios, which in turn causes the positioning results of drones to jump when crossing different micro-topographic areas, frequently triggering their obstacle avoidance or replanning systems.

[0004] A deeper analysis reveals that the aforementioned limitations stem from two aspects of the processing logic in existing technologies. First, purely mathematical interpolation logic cannot take real geographical barriers such as mountains or coastlines as physical constraints in the interpolation process, leading to the incorrect propagation of correction information to physically disconnected areas. This process does not conform to the true physical process of tropospheric water vapor distribution. Second, existing methods treat data sources with different timeliness and physical characteristics equally in a mathematical manner, thus failing to effectively remove systematic biases introduced by time delays in the background field. Consequently, the reliability of the final product is limited by the update frequency of the background field data. Therefore, how to establish a new technical approach that can break free from the assumption of a stable tropospheric field and, in the process of generating high spatiotemporal resolution tropospheric products, fully utilize the spatial coverage advantages of remote sensing satellites while accurately reflecting the blocking effects of complex terrain and the real-time dynamic changes of tropospheric delays becomes the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides a method for generating tropospheric delay products by combining GNSS and polar-orbiting remote sensing satellites. Its main purpose is to solve the problem that the existing static fusion method based on a posteriori weighting cannot generate physically reliable and spatiotemporally continuous high-precision tropospheric delay products when facing complex terrain and dynamic weather, because it relies on the inherent assumption of stable changes in tropospheric delay.

[0006] To achieve the above objectives, this invention provides a method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites, comprising the following steps: Step a: Obtain the background field of tropospheric delay from remote sensing satellites covering the target area, as well as the ground truth points of tropospheric delay acquired in real time from at least two GNSS stations within the target area. Step b: At the location of each GNSS station, calculate the difference between the true value of the tropospheric delay and the value of the background tropospheric delay field of the remote sensing satellite at the corresponding location to generate a set of differential anchor points. Step c: Perform reliability arbitration on a set of differential anchor points to identify and correct outlier anchor points; Step d: Based on the differential anchor points that have undergone reliability arbitration, a continuous tropospheric differential correction field covering the target area is generated through a terrain-constrained interpolation algorithm. When calculating the correction value at any location, the terrain-constrained interpolation algorithm uses the effective path distance as the basis for its weights. The effective path distance takes into account the terrain obstacle penalty. Step e involves overlaying the tropospheric differential correction field onto the remote sensing satellite tropospheric delay background field to generate the final tropospheric delay product.

[0007] Preferably, the terrain constraint interpolation algorithm is the inverse distance weighting method. The effective path distance is calculated based on digital elevation model data. If the path between two geographic points must cross an elevation obstacle identified by the digital elevation model data, the calculated distance of the path is increased by a penalty coefficient defined within the system.

[0008] Preferably, the reliability arbitration in step c includes: for each target differential anchor point to be arbitrated, temporarily removing it from a set of differential anchor points to form a subset of neighbor anchor points; using a terrain-constrained interpolation algorithm and based on the subset of neighbor anchor points, calculating a virtual differential value at the location of the target differential anchor point; comparing the real value of the target differential anchor point with the virtual differential value; if the absolute value of the difference between the two exceeds a statistical threshold determined based on the historical tropospheric delay statistical characteristics of the target area, then the target differential anchor point is determined to be an outlier anchor point and replaced with a virtual differential value.

[0009] Preferably, after step b, a step of identifying the physical cause state of the differential anchor points is further included. The identification step includes: at the GNSS station, synchronously acquiring the carrier phase high-frequency variance and the signal-to-noise ratio fluctuation variance of the received GNSS signal, and constructing an artifact identification index. ,in, The variance of the signal-to-noise ratio fluctuation. Let V be the carrier phase high-frequency variance, and if the artifact discrimination index is used during step d. If the threshold is exceeded, an inhibitory contribution weight is assigned to the differential anchor point; otherwise, a normal contribution weight is assigned.

[0010] Preferably, it also includes an adaptive processing mode switching mechanism based on the spatial density of differential anchor points. The mechanism includes: before executing step c, evaluating the spatial density of differential anchor points in the target area; if the spatial density is lower than a data abundance threshold, then suspending the execution of steps c to e, and instead executing the following temporal gradient feedforward step: based on historically archived remote sensing satellite tropospheric delay background field data, calculating the temporal evolution gradient of each grid point in the target area, using at least one differential anchor point in the target area or its neighboring area to perform overall calibration of the temporal evolution gradient field, and based on the calibrated temporal evolution gradient, extrapolating from the most recent historical remote sensing period to the current time to generate the final tropospheric delay product.

[0011] Preferably, the method further includes the step of generating an accompanying positioning quality index field. The steps include: defining the absolute value of the residual between the true value of each target differential anchor point and its corresponding virtual differential value, calculated in the reliability arbitration step, as the local tropospheric instability index of that location; generating a continuous positioning quality index field covering the target area based on the local tropospheric instability indices of all GNSS station locations and using a terrain-constrained interpolation algorithm; and outputting the positioning quality index field together with the final tropospheric delay product.

[0012] Preferably, elevation barriers are defined as continuous areas where the terrain slope or elevation change rate exceeds a barrier threshold.

[0013] Preferably, the statistical threshold is dynamically determined based on the historical tropospheric delay statistical characteristics of the target area and the average terrain path distance between GNSS stations in the area.

[0014] Preferably, a suppressive contribution weight is assigned to the differential anchor point, specifically by multiplying its weight in the inverse distance weighting method by a suppression factor less than 1. The value of the suppression factor is related to the artifact detection index. The magnitudes are positively correlated.

[0015] Preferably, the final output tropospheric delay product is a two-layer data packet, with the first data layer being the final tropospheric delay product and the second data layer being the positioning quality index field, for downstream high-precision positioning applications to conduct risk assessment and decision-making.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. By calculating the difference between real-time high-precision GNSS ground truth points and outdated remote sensing background fields to generate a set of differential anchor points, and generating a tropospheric differential correction field based on this set of differential anchor points, this method changes the existing approach of directly mathematically fusing two types of data with different timeliness and physical characteristics. Instead of focusing on the absolute values ​​of the two types of data, this method focuses on the deviation between them. This deviation precisely captures the sum of the real atmospheric changes since the remote sensing satellite passed over and the inherent deviation of the remote sensing model. Thus, the problem is transformed from numerical fitting to dynamic correction of physical deviations, making the final tropospheric delay product more consistent and reliable in a physical sense.

[0017] 2. In generating the tropospheric differential correction field, this invention employs a terrain-constrained interpolation algorithm. The interpolation weights are based on effective path distances that take into account terrain obstacle penalties, rather than simple straight-line distances. This approach ensures that the propagation of correction information respects the physical reality of discontinuous tropospheric water vapor distribution caused by real geographical barriers such as mountains or coastlines. It avoids the inaccurate interpolation results produced by traditional interpolation methods in complex terrain areas due to incorrect crossing of physical barriers. Especially in scenarios such as drone logistics delivery in hilly areas, it can provide more refined and accurate tropospheric delay correction for high-precision positioning, reducing problems such as decreased flight efficiency and increased energy consumption caused by positioning jumps.

[0018] 3. This invention establishes an internal data quality self-consistent verification mechanism that does not require an external reference source by performing reliability arbitration on differential anchor points before generating the correction field. This arbitration uses the idea of ​​leave-one-out cross-validation to construct a virtual truth value for each anchor point using its neighboring anchor points. By comparing the difference between the true value and the virtual truth value, outlier points caused by transient failures of GNSS stations are identified and corrected. This self-cleaning capability avoids local contamination of the correction field by erroneous differential anchor points, and improves the fault tolerance of the entire technical process to fluctuations in input data quality and the stability of the final product from a methodological perspective.

[0019] 4. This invention also provides a physical cause state discrimination mechanism. By simultaneously analyzing two accompanying parameters with different physical meanings—the high-frequency variance of the carrier phase and the variance of the signal-to-noise ratio fluctuation of the GNSS signal—it can distinguish whether the differential anchor point is caused by real water vapor activity or by non-water vapor factors such as ice crystal artifacts that cannot be distinguished by remote sensing detectors. This method based on multi-dimensional information collaborative judgment improves the system's ability to understand the physical causes behind remote sensing data, avoids incorrectly correcting remote sensing artifacts as real atmospheric changes, and thus prevents the introduction of systematic deviations that contradict physical reality into the final product.

[0020] 5. This invention reuses the correction residuals generated during the reliability arbitration process to generate an accompanying positioning quality index field in parallel. This design transforms the residual information, which was originally intermediate computational waste, into a second product with clear physical meaning that can characterize the local tropospheric spatiotemporal stability. This makes the final output no longer a single tropospheric delay value, but a decision-level information package that includes the value and its confidence assessment. This provides a decision-making basis for downstream risk-sensitive applications such as autonomous driving or drones to dynamically adjust their behavioral strategies. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall process of the tropospheric delayed product generation method of the present invention; Figure 2 This is the logical data flow diagram for multi-source data fusion and two-layer product generation in this invention; Figure 3 This is a diagram of the full-chain service architecture of the present invention, from data collection to terminal application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in detail below. However, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and such modifications and equivalent substitutions should all be covered within the protection scope of the present invention.

[0023] A method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites is disclosed. Its operation primarily includes data acquisition, differential anchor point generation and arbitration, terrain-constrained correction field construction, and final product synthesis and output. The data flow begins with the acquisition of multi-source data. A set of differential anchor points characterizing physical deviations is generated and arbitrated. A tropospheric differential correction field is constructed using an interpolation algorithm that considers geographical barriers. This correction field is then superimposed on the remote sensing background field to generate a corrected tropospheric delay product. In applications providing high-precision positioning services to users such as drones in terrain-complex areas like hilly or coastal regions, a key technical challenge stems from the tropospheric delay background field provided by polar-orbiting remote sensing satellites. Due to its time delay, it exhibits a systematic deviation from the current true atmospheric conditions. To address this challenge, this solution is configured as follows: within the target area, acquire the tropospheric delay background field covered by remote sensing satellites; simultaneously, acquire the true tropospheric delay points in real time at at least two GNSS stations deployed within the area, and at the location of each GNSS station... Perform differential calculations to generate differential anchor points. Its calculation formula is The generated differential anchor point set Its value represents the sum of the actual atmospheric changes and the inherent bias of the remote sensing model at each station location from the time the remote sensing satellite passes over to the present moment. Considering that a single GNSS station may generate instantaneous outliers due to local interference, thus affecting the accuracy of the correction field, the system executes a reliability arbitration procedure: for each target differential anchor point to be arbitrated in the set... First, logically remove the target anchor point from the anchor point set, forming a subset of neighboring anchor points. Then, call the terrain constraint interpolation algorithm (described later) to calculate the target differential anchor point based on this subset of neighboring anchor points. A virtual difference value at the location Next, the absolute value of the residual between the true value and the virtual difference value is calculated. The residual is then compared with a statistical threshold, which is dynamically determined based on the historical tropospheric delay statistics of the target area and the average terrain path distance between GNSS stations within the area. This threshold adapts to the atmospheric variation characteristics and station density in different regions. If the absolute value of the residual exceeds this dynamic threshold, the target differential anchor point is determined. Use outlier anchor points and their corresponding virtual difference values. The data is then replaced and used for subsequent correction field generation. To further distinguish whether the differential anchor points are caused by real water vapor changes or by remote sensing physical artifacts such as ice crystals in upper-level thin cirrus clouds, the system also performs physical cause state discrimination, synchronously acquiring the carrier phase high-frequency variance of the signal at each GNSS station. With signal-to-noise ratio fluctuation variance And construct an artifact detection index , calculate The value is compared with a discrimination threshold. If the threshold is exceeded, it indicates that the difference anchor point at that location is likely affected by non-moisture factors. Accordingly, in subsequent interpolation calculations, a suppressive contribution weight is assigned to this anchor point. This weight is obtained by multiplying its standard weight in the inverse distance weighting method by a factor less than 1 and proportional to... The system employs a positively correlated suppression factor to achieve this. After obtaining a set of arbitrated and screened differential anchor points, the system generates a continuous tropospheric differential correction field covering the target area using a terrain-constrained interpolation algorithm. The interpolation weights used in this algorithm are based on the effective path distance, rather than a simple straight-line distance. The calculation of the effective path distance is based on the digital elevation model (DEM) data. When calculating the path between any two points, if the path must traverse an identified elevation barrier or a continuous area where the terrain slope or elevation change rate exceeds a preset barrier threshold, then a penalty coefficient defined within the system will be applied to the calculated path distance. This method ensures that the spatial propagation of correction information follows the constraints of physical barriers.

[0024] By superimposing the generated tropospheric differential correction field onto the original remote sensing satellite tropospheric delay background field, the final tropospheric delay product is generated. Simultaneously, the system utilizes the absolute value of the residuals calculated in the reliability arbitration step, i.e., the local tropospheric instability index, based on this index for all GNSS station locations, and employs the same terrain-constrained interpolation algorithm to generate a continuous positioning quality index field in parallel. The output product is a two-layer data packet: data layer one is the final tropospheric delay product, and data layer two is the positioning quality index field, for downstream applications to perform risk assessment and decision-making. In scenarios where SS sites are sparsely distributed, the system is configured with a processing mode switching mechanism. Before generating the correction field, the spatial density of the differential anchor points is evaluated. If the density is lower than the preset data abundance threshold, the spatial interpolation step is stopped, and the temporal gradient feedforward step is executed instead: based on the historical archived remote sensing background field data, the temporal evolution gradient of each grid point in the target area is calculated, and the gradient field is calibrated as a whole using at least one differential anchor point in the area or its vicinity. Based on the calibrated gradient, the current moment is extrapolated from the most recent historical remote sensing period to generate a tropospheric delay product.

[0025] System adjustment coefficient in the physical origin state discrimination mechanism The specific values ​​are set according to a set of optimized calibration procedures based on historical data. This procedure first constructs an independent validation dataset containing known classification labels, namely clean weather samples and artifact samples, from historical GNSS signal data and meteorological data from the same period. Then, a comprehensive index is introduced to evaluate the classification performance. Fractions, which are defined as Among them, accuracy The proportion of correctly identified artifact samples out of all samples identified as artifacts is denoted as Recall, which is the proportion of correctly identified artifact samples out of all real artifact samples. Then, within a pre-defined parameter scanning interval, different... The value corresponding to The score, and will eventually make The score reaches the maximum value The numerical values ​​are determined as the final engineering parameters for deployment in this area; the data abundance threshold required for the adaptive processing mode switching mechanism and the positive margin in the terrain-constrained interpolation algorithm are also considered. Its calibration process is directly related to the service level requirements and basic data attributes of specific application scenarios.

[0026] Example 1: In a continuously operating automated logistics delivery operation using unmanned aerial vehicles (UAVs), the flight path needs to cross a ridge with a significant elevation difference from a distribution center located in a valley to reach a designated location on the other side. The positioning correction service used for this flight path relies on a conventional tropospheric delay product that combines ground GNSS station data and remote sensing satellite data. When the UAV crosses the ridge peak along the predetermined path, the three-dimensional coordinates calculated by its onboard positioning system experience a momentary jump. This jump triggers the obstacle avoidance and flight path replanning procedures of the flight control system, causing the aircraft to hover and perform energy-intensive attitude adjustments. Ultimately, this results in the delivery mission failing to meet timeliness standards and the return battery power falling below the warning value. To address this situation, the tropospheric delay product generation method of this invention is deployed in the operating area. The system acquires the remote sensing satellite tropospheric delay background field at the same time as the aforementioned scenario. and real-time tropospheric delay truth points of several GNSS stations deployed on the plains on the other side of the valley and ridge. Then, a set of differential anchor points was first generated. In terms of technical processing, this step no longer performs direct mathematical fitting of the absolute numerical fields of the two heterogeneous sources, but instead turns to solving a physical difference field that characterizes the systematic deviation of the background field. Subsequently, the system performs reliability arbitration on the group of difference anchor points, identifies and corrects the erroneous anchor point corresponding to one of the GNSS stations that has generated outliers due to local signal interference.

[0027] In the process of generating the tropospheric differential correction field, the two technical features of differential anchor point generation and terrain-constrained interpolation algorithm are interdependent. The former provides the correction information to be propagated with clear physical meaning for the interpolation process, while the latter provides the propagation path of this information in accordance with physical laws. In specific operation, when calculating the correction value at the ridge peak, the interpolation algorithm, because its weight is based on the effective path distance that takes into account the terrain obstacle penalty, causes the weight contribution of the differential anchor point located in the valley to the ridge peak to be greatly reduced because the effective path distance between the two points needs to detour around the ridge. The correction value at this location is mainly determined by the contribution of the differential anchor point on the plain side. This mechanism not only utilizes the spatial coverage of the remote sensing background field, but also achieves physical correction of local details through the combination of GNSS anchor points and terrain-constrained interpolation. When the UAV uses the tropospheric delay product generated by this invention to perform the same flight path mission again, its positioning solution remains continuous throughout the entire flight profile. During the flight over the ridge, the positioning coordinates do not jump, the flight attitude is stable, and its final mission flight time and energy consumption meet the expectations of standard operation.

[0028] Example 2: To verify the effectiveness of the technical solution of the present invention in practical applications, especially its effect on improving the accuracy of tropospheric delay products in areas with complex terrain, this comparative experiment was designed and executed. The purpose of the experiment is to introduce a control group using conventional interpolation methods in the industry, and under the same input data conditions, quantitatively evaluate the accuracy performance of the method of the present invention compared with the conventional method at different terrain feature points. The test platform was selected in a coastal mountain area, which simultaneously has mountains with drastic elevation changes and relatively flat coastal plains, and is equipped with a high-precision GNSS reference station network that can provide continuous observation data. The test data source includes the tropospheric delay background field covering the entire test area, acquired by a polar-orbiting remote sensing satellite. Real-time tropospheric delay truth points obtained from the GNSS reference station network in the test area In addition, a 30-meter resolution digital elevation model covering the area was used to construct the experimental and control groups. A subset of stations from the GNSS network were selected as the basic data source for generating differential anchor points, while the remaining stations served as independent accuracy check points. The value is regarded as the reference true value at that location and does not participate in the product generation process of the two subsequent methods; the experimental group adopts the technical solution method A of the present invention, while the control group adopts a weighted average interpolation scheme method B based on straight-line distance. Both schemes use the same remote sensing background field and differential anchor point basic data source.

[0029] Regarding the experimental parameter settings, for the terrain-constrained interpolation algorithm in Scheme A of this invention, the key parameter, the terrain obstacle penalty coefficient, is set to balance effective isolation of physical barriers with smooth adaptation to general terrain undulations. If the parameter value is too small, it will not be able to prevent the correction information from crossing major mountain ranges; if the value is too large, it may generate unnecessary gradient changes on secondary terrain. The setting procedure for this coefficient is to ensure that the path penalty value applied is numerically greater than the maximum tropospheric delay difference that may occur when crossing major terrain obstacles, as statistically derived from regional historical data. In the coastal mountain scenario of this experiment, this coefficient is set as a non-limiting example value. After the experiment started, Method A and Method B generated tropospheric delay products covering the entire experimental area in parallel during a continuous 48-hour experimental period. At each independent accuracy checkpoint, the tropospheric delay values ​​generated by the two methods were compared with the reference true value. At the relatively flat coastal plain checkpoint CK-01, the root mean square errors of Method A and Method B were 4.1 mm and 5.3 mm, respectively, showing similar performance. However, in areas with rapidly changing terrain features... The accuracy differences between the two methods are significant at checkpoints with dramatic variations. At checkpoint CK-04, located on a ridgeline, the root mean square error of method B increases to 25.4 mm, while the error of method A is 7.1 mm. Similarly, at checkpoint CK-03 inside the valley, the error of method B is 18.5 mm, while the error of method A is 6.2 mm. This difference in error stems from the fact that method B uses interpolation logic based on straight-line distance, which performs a cross-terrain mathematical average of the differential anchor point information on both sides of the valley and the plain. In contrast, the terrain-constrained interpolation algorithm in method A, due to its effective path distance design, prevents the propagation of information across physical barriers, thus obtaining a correction result that is more consistent with the physical process. Experimental data shows that the technical solution provided by this invention, under the condition of using the same input data source as a conventional solution, generates a tropospheric delay product with smaller errors in complex terrain areas. This reduction in error comes from the fact that its terrain-constrained interpolation algorithm follows the physical laws of water vapor distribution, thereby avoiding the model distortion problem caused by the conventional solution's inability to take into account real geographical barriers.

[0030] Example 3: This example combines Figures 1 to 3 The method for generating tropospheric delay products from joint GNSS and polar-orbiting remote sensing satellites is explained, such as... Figure 1As shown, the process begins with data acquisition, specifically obtaining the tropospheric delay ground truth points from real-time GNSS station data and the tropospheric delay background field from remote sensing satellite data. Then, a differential anchor point generation step calculates the deviation between the GNSS ground truth and the remote sensing background field, and performs reliability arbitration on the differential anchor points to identify and correct outlier anchor points. The correction residuals output from this arbitration step are used to generate the accompanying positioning quality index. The arbitrated anchor points, supported by digital elevation model (DEM) data, generate correction information through a tropospheric differential correction field generation step and apply a terrain-constrained interpolation algorithm. The generated correction field is then superimposed onto the original background field through a product synthesis step, forming the final output two-layer data packet together with the accompanying positioning quality index field. This packet contains data layer one: the final tropospheric delay product and data layer two: the positioning quality index field.

[0031] like Figure 2 As shown, the external entity, the GNSS network, provides the tropospheric delay ground truth points, while the external entity, the remote sensing satellite system, provides the tropospheric delay background field. Data from both are fed into module 1.0 to generate differential anchor points, producing an initial set of differential anchor points. This set is then processed by module 2.0 for reliability arbitration, generating arbitrated differential anchor points and correction residuals. The former, along with elevation data from data storage D1 (Digital Elevation Model DEM), is input into module 3.0 to generate a differential correction field. The latter, along with the elevation data, is input into module 4.0 to generate a positioning quality index field. The tropospheric differential correction field and positioning quality index field produced by modules 3.0 and 4.0, together with the initial tropospheric delay background field provided by the remote sensing satellite system, are finally synthesized into a two-layer data packet in module 5.0 and delivered to the external entity: downstream high-precision positioning applications.

[0032] like Figure 3 As shown, the node polar-orbiting remote sensing satellite provides remote sensing background field data through the satellite downlink, while the component real-time data acquisition module in the node GNSS ground station network provides GNSS delay truth points through the ground network. These two data streams are converged to the node data and algorithm processing center, where they are processed by its internal component tropospheric delay product generation system to generate tropospheric delay products. These products are then provided to node user terminals, such as UAVs, through data broadcasting services. The component positioning correction service client in the user terminal receives these products for positioning correction.

[0033] Example 4: Before deploying the method of this invention in a new operating area, to match the key parameters of its internal algorithm with the tropospheric and topographical environmental characteristics of the area, a standardized offline calibration procedure is executed to provide a set of initial values ​​with engineering setting basis for the automated operation of the system. Regarding the statistical threshold in the reliability arbitration mechanism, the calibration process is as follows: Obtain the historical tropospheric delay ground truth time series data of all available GNSS stations in the target operating area for at least one year, and the digital elevation model data of the area. For any station in the network, traverse all historical moments, using other stations at that moment as a subset of neighbor anchor points, and use a terrain-constrained interpolation algorithm to calculate a virtual difference value sequence at the target station location. Subsequently, the difference between the station's true value sequence and the virtual value sequence is calculated to obtain a residual sequence, and the root mean square value of the residual sequence is calculated. Iterate through all stations in the network and repeat the above steps until all stations have calculated their scores. The average value is taken, and three times that value is set as the statistical threshold for the region, which is used to determine outlier anchor points during online operation.

[0034] Artifact discrimination index in physical origin state discrimination mechanism The required discrimination threshold and inhibitory contribution weight are determined as follows: Historical GNSS signal data of the target area are collected, and combined with meteorological satellite cloud imagery data from the same period, two types of sample datasets are selected. One type consists of samples where tropospheric delay changes are dominated by water vapor activity under clear weather conditions, and the other type consists of samples with clear upper-level coverage by thin cirrus clouds composed of ice crystals. The corresponding values ​​for these two types of sample datasets are calculated respectively. The exponential distribution, where the discrimination threshold is set at the numerical point that best distinguishes the two distributions, is used for calculating the suppressive contribution weight when a certain difference anchor point... When the index exceeds the screening threshold, its weight in the interpolation calculation will be multiplied by a suppression factor. This factor follows the formula ,in, This is a system adjustment coefficient greater than 0, used to control the rate at which the weight decreases.

[0035] The calibration procedure for the terrain obstacle threshold and penalty coefficient required by the terrain constraint interpolation algorithm is as follows: Based on the digital elevation model data of the target area, calculate the slope distribution histogram of the entire area, and set the slope value at the 95th percentile of this distribution as the terrain obstacle threshold. This identifies the main physical barriers such as ridgelines in the area. Subsequently, for a typical terrain obstacle identified by this threshold, calculate the average straight-line path length for the machine to traverse the obstacle. And the average shortest effective path distance around the obstacle. To force the algorithm to choose a detour during pathfinding, a penalty coefficient is applied. The value is set to ,in, It provides a small positive margin; by executing the above procedures, before deployment, all key parameters of the method of the present invention are given quantitative values ​​that match the environmental characteristics of the specific work area and have a clear basis for setting.

[0036] Example 5: When the method of the present invention is applied to a large area with uneven GNSS station distribution density, in order to calibrate the data abundance threshold on which its internal adaptive processing mode switching mechanism is based, a simulation analysis procedure needs to be executed in advance. This procedure first selects the sub-region with the most dense GNSS station distribution in a region, and uses all the station data in the sub-region to generate a high-precision tropospheric delay reference field. Subsequently, the system enters a simulation iteration loop. In each iteration, an anchor point is removed from the differential anchor point set used to generate the correction field in a predetermined sequence to simulate the gradual decrease in station density.

[0037] After each iteration, the system regenerates a test field based on the remaining set of anchor points using terrain-constrained interpolation. It then calculates the root mean square error (RMSE) of this test field relative to the aforementioned high-precision reference field. This process continues iteratively until the number of anchor points is reduced to a preset minimum. The system records complete data on the change in RMSE with anchor point spatial density. The point where the error growth rate first exceeds a preset threshold is identified as a performance inflection point, and the spatial density value corresponding to this inflection point is set as the data abundance threshold. This procedure directly links the triggering condition for mode switching to a quantifiable rate of change in output product accuracy. The determination of the data abundance threshold begins with a clearly defined application scenario and a tolerable error increment. This increment is a performance indicator pre-set based on downstream risk control strategies such as those for drone autonomous navigation. Root mean square error of the reference field Calculate a root mean square error threshold for performance alarms. During the simulation iteration process with gradually decreasing site density, the root mean square error of the test field was first achieved. The spatial density of anchor points corresponding to the time is directly determined as the data abundance threshold, while the positive margin in the terrain obstacle penalty is also considered. This is related to the raster resolution of the digital elevation model (DEM). The values ​​are bound together and are deterministically set to... This provides a decision bias based on the physical properties of the data source and non-arbitrary factors for path optimization.

[0038] Example 6: To ensure that the service level of the method of the present invention can be clearly defined and guaranteed in applications with uninterrupted and high reliability requirements, a set of fault-tolerant and service degradation operation procedures covering the main data source interruption scenarios needs to be configured before the system is officially put into operation. The procedures define the logical path to be executed when the system detects that the upstream data link is interrupted, resulting in the complete loss of key input data streams within a predetermined update cycle.

[0039] The procedure includes the following: when the system detects a new tropospheric delay background field from remote sensing satellites. If the arrival does not occur within the time window set based on its nominal update frequency, the system will maintain service and use the last successfully received background field as a reference. Simultaneously, a temporal compensation is introduced: based on the average change trend of all real-time differential anchors since the last valid background field, an overall temporal deviation correction is applied to the outdated background field. This correction is then superimposed on the subsequently generated spatial correction field. When the system detects an interruption in the real-time data stream of all GNSS stations, resulting in the inability to generate any valid differential anchors, this state is determined to be due to data abundance falling below the previously specified data abundance threshold. The system then automatically switches to temporal gradient feedforward mode, retrieves the most recent historical remote sensing periodic data, and applies the specified temporal evolution gradient for extrapolation. During the execution of any fault-tolerant procedure, the system fills the accompanying positioning quality index field in its final generated two-layer data packet with a preset value representing low confidence and attaches a specific data quality status flag to provide downstream applications with clear information about the current product service status.

[0040] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites, characterized in that, Includes the following steps: Step a: Obtain the background field of tropospheric delay from remote sensing satellites covering the target area, and the true tropospheric delay points acquired in real time from at least two GNSS stations within the target area; Step b: At the location of each GNSS station, calculate the difference between the true value of the tropospheric delay and the value of the background tropospheric delay field of the remote sensing satellite at the corresponding location to generate a set of differential anchor points. Step c: Perform reliability arbitration on a set of differential anchor points to identify and correct outlier anchor points; Step d: Based on the differential anchor points that have undergone reliability arbitration, a continuous tropospheric differential correction field covering the target area is generated through a terrain-constrained interpolation algorithm. When calculating the correction value at any location, the terrain-constrained interpolation algorithm uses the effective path distance as the basis for its weights. The effective path distance takes into account the terrain obstacle penalty. Step e involves overlaying the tropospheric differential correction field onto the remote sensing satellite tropospheric delay background field to generate the final tropospheric delay product.

2. The method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites according to claim 1, characterized in that, The terrain-constrained interpolation algorithm is an inverse distance weighted method. The effective path distance is calculated based on digital elevation model data. Between two geographic points, if the path must cross an elevation obstacle identified by the digital elevation model data, the calculated distance of the path is increased by a penalty coefficient defined within the system.

3. The method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites according to claim 1, characterized in that, The reliability arbitration in step c includes: for each target differential anchor point to be arbitrated, temporarily removing it from a set of differential anchor points to form a subset of neighbor anchor points; using a terrain-constrained interpolation algorithm and based on the subset of neighbor anchor points, calculating a virtual differential value at the location of the target differential anchor point; comparing the real value of the target differential anchor point with the virtual differential value; if the absolute value of the difference between the two exceeds a statistical threshold determined based on the historical tropospheric delay statistical characteristics of the target area, then the target differential anchor point is determined to be an outlier anchor point and replaced with a virtual differential value.

4. The method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites according to claim 1, characterized in that, Following step b, the method further includes a step of identifying the physical cause state of the differential anchor points. This identification step includes: at the GNSS station, synchronously acquiring the carrier phase high-frequency variance and signal-to-noise ratio fluctuation variance of the received GNSS signal, and constructing an artifact identification index. ,in, The variance of the signal-to-noise ratio fluctuation. Let V be the carrier phase high-frequency variance, and if the artifact discrimination index is used during step d. If the threshold is exceeded, an inhibitory contribution weight is assigned to the differential anchor point; otherwise, a normal contribution weight is assigned.

5. The method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites according to claim 1, characterized in that, It also includes an adaptive processing mode switching mechanism based on the spatial density of differential anchor points. The mechanism includes: before executing step c, the spatial density of differential anchor points in the target area is evaluated. If the spatial density is lower than a data abundance threshold, the execution of steps c to e is stopped, and the following temporal gradient feedforward steps are executed instead: based on the historical archived remote sensing satellite tropospheric delay background field data, the temporal evolution gradient of each grid point in the target area is calculated. The temporal evolution gradient field is calibrated as a whole using at least one differential anchor point in the target area or its neighboring area. Based on the calibrated temporal evolution gradient, the time is extrapolated from the most recent historical remote sensing period to the current time to generate the final tropospheric delay product.

6. The method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites according to claim 3, characterized in that, It also includes the step of generating an accompanying positioning quality index field, which includes: defining the absolute value of the residual between the true value and the corresponding virtual differential value of each target differential anchor point calculated in the reliability arbitration step as the local tropospheric instability index of that location; generating a continuous positioning quality index field covering the target area based on the local tropospheric instability indices of all GNSS station locations and using a terrain-constrained interpolation algorithm; and outputting the positioning quality index field together with the final tropospheric delay product.

7. The method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites according to claim 2, characterized in that, Elevation barriers are defined as continuous areas where the slope or rate of change of elevation exceeds a barrier threshold.

8. The method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites according to claim 3, characterized in that, The statistical threshold is dynamically determined based on the historical tropospheric delay statistical characteristics of the target area and the average terrain path distance between GNSS stations in the area.

9. The method for generating tropospheric delay products using a combination of GNSS and polar-orbiting remote sensing satellites according to claim 4, characterized in that, Assign a suppressive contribution weight to this difference anchor point, specifically by multiplying its weight in the inverse distance weighting method by a suppression factor less than 1. The value of the suppression factor is related to the artifact detection index. The magnitudes are positively correlated.

Images