Zenith troposphere delay prediction method, zenith troposphere delay prediction device and electronic equipment

Abstract

The embodiment of the invention provides a zenith troposphere delay prediction method, a zenith troposphere delay prediction device and electronic equipment. The method comprises the following steps: acquiring multi-source heterogeneous data; performing alignment on the multi-source heterogeneous data to obtain global grid data; according to the global grid data, zenith statics delay ZHD and zenith wet delay ZWD are modeled to obtain a ZHD long-period prediction model and a ZWD short-time prediction model; according to the global grid data, constructing a correction model which changes along with the elevation in the vertical direction; and according to the ZHD long-period prediction model, the ZWD short-time prediction model and the correction model, performing ZTD prediction on a to-be-measured station. And the prediction accuracy of the ZTD is improved, so that support is provided for improving the stability and precision of GNSS positioning.

Description

Technical Field

[0001] This application relates to the field of satellite navigation and positioning technology, and in particular to a method, apparatus and electronic device for predicting zenith tropospheric delay. Background Technology

[0002] With the widespread application of high-precision Global Navigation Satellite System (GNSS) in various fields, tropospheric error has become one of the key factors restricting positioning accuracy. Zenith Tropospheric Delay (ZTD) is mainly caused by the distribution of water vapor and changes in air pressure in the atmosphere. Its spatiotemporal variation characteristics are significant, especially in complex terrain areas such as mountains, valleys, and plateaus. The elevation difference caused by terrain undulations will exacerbate the non-uniformity of zenith tropospheric delay, thereby affecting the stability and accuracy of GNSS positioning. Summary of the Invention

[0003] This application provides a method, apparatus, and electronic device for zenith tropospheric delay prediction, which improves the prediction accuracy of ZTD and thus provides support for improving the stability and accuracy of GNSS positioning.

[0004] In a first aspect, embodiments of this application provide a method for predicting zenith tropospheric delay, including:

[0005] Acquire multi-source heterogeneous data;

[0006] The multi-source heterogeneous data is aligned to obtain global grid data;

[0007] Based on the global grid data, the zenith static delay (ZHD) and zenith wet delay (ZWD) are modeled to obtain a long-term prediction model for ZHD and a short-term prediction model for ZWD.

[0008] Based on the global grid data, a modified model is constructed to address the vertical variation with elevation.

[0009] Based on the ZHD long-term prediction model, the ZWD short-term prediction model, and the modified model, ZTD prediction is performed on the site to be tested.

[0010] In one possible implementation, the multi-source heterogeneous data includes Global Navigation Satellite System (GNSS) observation data and station meteorological data;

[0011] After acquiring the multi-source heterogeneous data, the method further includes:

[0012] Precise point positioning is performed based on the GNSS observation data to obtain the ZTD of each station;

[0013] Based on the meteorological data of the stations and the Sastamonin model, the ZHD of each station was determined;

[0014] For each site, the difference between the ZTD and ZHD of the site is determined as ZWD;

[0015] The ZHD and ZWD of each station were determined as the preprocessed GNSS observation data.

[0016] In one possible implementation, the multi-source heterogeneous data includes meteorological profile data, GNSS observation data, and reanalysis meteorological data;

[0017] The process of aligning the multi-source heterogeneous data to obtain global grid data includes:

[0018] The GNSS observation data and reanalysis meteorological data are processed to achieve unified time resolution.

[0019] The meteorological profile data is matched with the time-resolution unified GNSS observation data and reanalysis meteorological data through time window matching, and then aligned using an interpolation algorithm to obtain global grid data with unified time resolution.

[0020] In one possible implementation, the step of modeling the zenith static delay (ZHD) and zenith wet delay (ZWD) based on the global grid data to obtain a long-term prediction model for ZHD and a short-term prediction model for ZWD includes:

[0021] Based on the reanalysis meteorological data and GNSS observation data in the global grid data, ZHD is modeled to obtain a long-term prediction model for ZHD.

[0022] Based on the reanalysis meteorological data and meteorological profile data in the global grid data, ZWD is modeled to obtain a short-term ZWD prediction model.

[0023] In one possible implementation, the step of modeling ZHD based on reanalysis meteorological data and GNSS observation data from the global grid data to obtain a long-term ZHD prediction model includes:

[0024] Meteorological data of multiple pressure layers at each point are obtained from the reanalysis meteorological data in the global grid data; the meteorological data includes air pressure, potential energy and temperature.

[0025] For each site, the surface layer ZHD of the site is obtained by integrating the meteorological time series data of the multi-pressure layer of the site along the path;

[0026] Using spatial information as a weighting factor, the ZHD of each station in the GNSS observation data in the global grid data is fused with the surface ZHD of each point in the reanalysis meteorological data through an interpolation algorithm to obtain surface ZHD fused grid data.

[0027] The frequency of periodic changes is determined based on the periodic variation characteristics in the ZHD fused grid data of the surface layer;

[0028] A long-period prediction model for ZHD is constructed using a Fourier series fitting model, based on the ZHD fused grid data of the surface layer and the periodic variation frequency.

[0029] In one possible implementation, the step of modeling ZWD based on reanalysis meteorological data and meteorological profile data from the global grid data to obtain a short-term ZWD prediction model includes:

[0030] Multiple pressure layers are selected as standard pressure layers from the reanalysis meteorological data. The data of each pressure layer in the meteorological profile data are resampled to the standard pressure layers using an interpolation algorithm to obtain standardized meteorological profile data.

[0031] An attention-weighted mechanism is used to weight and fuse the standardized meteorological profile data and the reanalysis meteorological data to obtain fused meteorological grid data.

[0032] For each point in the meteorological grid data, the path integral in the vertical direction is performed based on the meteorological data of the point to obtain the ZWD time series data of the point;

[0033] For each point in the meteorological grid data, a Gaussian function is used to construct a short-term ZWD prediction model corresponding to the point based on the ZWD time series data of the point.

[0034] In one possible implementation, constructing a correction model for vertical elevation variations based on the global grid data includes:

[0035] Data fusion is performed based on the global grid data to obtain meteorological grid data, surface ZHD fused grid data, and ZWD fused grid data;

[0036] Using a neural network algorithm, a correction model for vertical elevation variations is constructed based on the meteorological grid data, surface ZHD fused grid data, and ZWD fused grid data.

[0037] In one possible implementation, the step of performing ZTD prediction on the site to be tested based on the ZHD long-term prediction model, the ZWD short-term prediction model, and the modified model includes:

[0038] Obtain the spatiotemporal information of the site to be tested; the spatiotemporal information is used to characterize the geographical location and the time of the test.

[0039] The spatiotemporal information is input into the ZHD long-period prediction model to obtain the ZHD prediction value of the site to be tested.

[0040] The spatiotemporal information is input into the ZWD short-time prediction model to obtain the ZWD prediction value of the site under test;

[0041] The spatiotemporal information is input into the correction model to obtain the correction value;

[0042] The ZTD prediction value of the site to be tested is determined based on the ZHD prediction value, ZWD prediction value, and the correction value.

[0043] Secondly, embodiments of this application provide a zenith tropospheric delay prediction device, comprising:

[0044] The acquisition module is used to acquire heterogeneous data from multiple sources.

[0045] An alignment module is used to align the multi-source heterogeneous data to obtain global grid data;

[0046] The modeling module is used to model the zenith static delay (ZHD) and zenith wet delay (ZWD) based on the global grid data, so as to obtain a long-term prediction model for ZHD and a short-term prediction model for ZWD.

[0047] The modeling module is also used to construct a corrected model for vertical changes with elevation based on the global grid data;

[0048] The prediction module is used to perform ZTD prediction on the site to be tested based on the ZHD long-term prediction model, the ZWD short-term prediction model, and the correction model.

[0049] Thirdly, embodiments of this application provide an electronic device, including: a processor and a memory communicatively connected to the processor;

[0050] The memory stores computer-executed instructions;

[0051] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0052] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0053] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0054] The zenith tropospheric delay prediction method, apparatus, and electronic equipment provided in this application, combined with multi-source heterogeneous data, design differentiated combined prediction strategies for the spatiotemporal characteristics of zenith tropospheric delay, such as the periodic variation of ZHD and the random variation of ZWD. By combining a long-period prediction model for ZHD and a short-term prediction model for ZWD, high-precision, real-time prediction of zenith tropospheric delay is achieved. Furthermore, by using a correction model for vertical elevation variations to address the impact of terrain elevation differences on modeling, the prediction accuracy of zenith tropospheric delay is improved, thereby providing support for improving the stability and accuracy of GNSS positioning. Attached Figure Description

[0055] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0056] Figure 1 A flowchart illustrating the ZTD prediction method provided in this application;

[0057] Figure 2 A schematic diagram of the overall ZTD prediction process provided in this application;

[0058] Figure 3 A schematic diagram of the ZTD prediction device provided in this application;

[0059] Figure 4 A schematic diagram of the structure of the electronic device provided in this application.

[0060] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0061] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0062] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0063] Figure 1 A flowchart illustrating the ZTD prediction method provided in this application is shown below. Figure 1 As shown, the method includes:

[0064] S101. Obtain multi-source heterogeneous data.

[0065] Multi-source heterogeneous data includes GNSS observation data, station meteorological data, reanalysis meteorological data, and meteorological profile data.

[0066] GNSS observation data includes observation data from each station at each sampling time over several consecutive years. This observation data includes pseudorange observation data, carrier phase observation data, and Doppler observation data.

[0067] Optionally, GNSS observation data may also include satellite ephemeris data, receiver clock bias and positioning information, and metadata. For example, metadata may include observation timestamps, station information, and satellite information.

[0068] The meteorological data from the stations include meteorological parameters measured at each sampling time over several consecutive years, specifically including air pressure, temperature, and relative humidity.

[0069] Reanalysis meteorological data refers to global reanalysis meteorological datasets, such as the fifth-generation global atmospheric reanalysis data (ERA5) provided by the European Centre for Medium-Range Weather Forecasts Reanalysis (ECMWF) and the reanalysis meteorological datasets provided by the National Centers for Environmental Prediction (NCEP) in the United States.

[0070] The meteorological profile data includes low-Earth orbit (LEO) atmospheric profile data and radiosonde meteorological profile data. The LEO atmospheric profile data is obtained by inverting radio occultation (RO) event data from GNSS receivers on LEO satellites.

[0071] S102. Align the multi-source heterogeneous data to obtain global grid data.

[0072] Because multi-source heterogeneous data have inconsistent time resolutions, and because low-Earth orbit atmospheric profile data is generated event-driven rather than obtained through timed sampling, it is necessary to perform time alignment on multi-source heterogeneous data to obtain global grid data.

[0073] The global grid data includes GNSS observation data with unified time resolution, station meteorological data, reanalysis meteorological data, and meteorological profile data.

[0074] Optionally, spatial alignment of multi-source heterogeneous data can be further performed. For example, stations in GNSS observation data can be aligned with network points in reanalysis meteorological data.

[0075] S103. Based on global grid data, model ZHD and ZWD respectively to obtain a long-term prediction model for ZHD and a short-term prediction model for ZWD.

[0076] The Zenith Tropospheric Delay (ZTD) includes the Zenith Hydrostatic Delay (ZHD) and the Zenith Wet Delay (ZWD).

[0077] The ZHD long-term prediction model is obtained by modeling the periodic characteristics of ZHD, while the ZWD short-term prediction model is obtained by modeling the short-term time series of ZWD. By modeling ZHD and ZWD separately, the modeling accuracy can be improved.

[0078] S104. Construct a corrected model for vertical changes in elevation based on global grid data.

[0079] Since the starting surface for reanalysis meteorological data is the surface or the elevation corresponding to 1000 hPa, which differs from the actual station elevation, and the water vapor variation near the surface is an important source of error affecting the zenith tropospheric delay modeling, this embodiment also constructs a correction model that varies with elevation in the vertical direction.

[0080] S105. Based on the ZHD long-term prediction model, ZWD short-term prediction model and correction model, perform ZTD prediction for the site to be tested.

[0081] The long-term ZHD prediction model is used to predict the ZHD of a site, the short-term ZWD prediction model is used to predict the ZWD of a site, and the correction model is used to correct the predicted ZHD and ZWD, thereby obtaining the accurate ZTD prediction value of the site to be tested.

[0082] In addition to commonly used tropospheric data sources such as "reanalysis meteorological data, GNSS observation data, and station meteorological data," this embodiment also incorporates Radiosonde meteorological profile data, which has higher accuracy than reanalysis meteorological data, as well as low-orbit atmospheric profile data with data utilization potential. It integrates multi-source heterogeneous data for three-dimensional modeling, and leverages the advantage of low-orbit atmospheric profile data supplementing the upper-level profile to improve the modeling error of reanalysis meteorological data at high levels. This enhances the tropospheric modeling capability for remote areas and sparsely populated station areas, solves the problem of GNSS observation blind spots, and is a universal modeling scheme for modeling all regions.

[0083] Based on the above embodiments, in an optional embodiment, after acquiring multi-source heterogeneous data, different preprocessing operations are required due to the spatiotemporal differences in the multi-source heterogeneous data.

[0084] The preprocessing of GNSS observation data includes: performing Precise Point Positioning (PPP) based on GNSS observation data to obtain the ZTD of each station; determining the ZHD of each station based on station meteorological data and the Saastamonien model; for each station, determining the difference between the ZTD and ZHD as the ZWD; and determining the ZHD and ZWD of each station as the preprocessed GNSS observation data.

[0085] The formula corresponding to the Saastamonien model is as follows:

[0086]

[0087]

[0088] in, For the site's ZHD, Surface air pressure, This is a latitude-elevation correction factor, used to adjust surface air pressure. Corrections are made to offset the effects of latitude and elevation on atmospheric statics, ensuring the accuracy of ZHD calculations. Higher elevations result in thinner atmospheres and smaller static delays, an effect quantified by a linear term (-0.00000028 × H). The latitude of the station is the geographical coordinate, ranging from -90° (South Pole) to 90° (North Pole), in degrees (°). The geodetic elevation of the station is expressed in meters (m), which refers to the vertical height of the station relative to the geoid.

[0089] This embodiment calculates the ZTD and ZHD of a site separately, and then calculates the ZWD based on these, thus achieving the effect of separating the ZHD and ZWD of a site.

[0090] The preprocessing of meteorological profile data includes: screening and quality control of low-orbit atmospheric profile data and Radiosonde meteorological profile data to obtain surface layer grid data and pressure layer grid data. Surface layer grid data includes surface specific humidity and surface temperature. Pressure layer grid data includes specific humidity and temperature for each pressure layer. For example, the screening and quality control are implemented by deleting data from the meteorological profile data that are incomplete or do not meet the profile height requirements (e.g., top pressure > 300 hPa), and then using a 3sigma outlier detection algorithm to remove outliers.

[0091] The preprocessing of meteorological data for reanalysis includes selecting data for a preset time span. For example, the preset time span may be several consecutive years, such as 3, 4, or 5 years.

[0092] The meteorological data analysis further includes surface layer grid data and barometric layer grid data. The surface layer grid data is defined as the near-surface layer (usually around 2 meters high, consistent with the meteorological observation altitude of GNSS stations). Surface layer grid data includes surface specific humidity, surface temperature, and surface air pressure. Barometric layer grid data includes specific humidity, temperature, and air pressure for each barometric layer. The barometric layer grid data is defined as vertical layers divided according to atmospheric pressure values ​​(e.g., 1000 hPa, 850 hPa, 700 hPa, etc., with lower pressure corresponding to higher altitudes).

[0093] The essence of a grid is to divide the Earth's surface and upper atmosphere into a regular grid at fixed latitude and longitude intervals (such as 0.25°×0.25°, 1°×1°). Each grid node (also called a grid point) corresponds to a unique latitude and longitude coordinate, and the grid node stores the meteorological parameter values ​​of the location indicated by the latitude and longitude coordinate, forming a gridded dataset.

[0094] Optionally, the multi-source heterogeneous data can be aligned after preprocessing.

[0095] In one optional embodiment, step S102, namely aligning multi-source heterogeneous data to obtain global grid data, includes: unifying the time resolution of GNSS observation data and reanalysis meteorological data; matching the meteorological profile data with the time-resolution unified GNSS observation data and reanalysis meteorological data through time windows, and aligning them through an interpolation algorithm to obtain global grid data with unified time resolution.

[0096] The time resolution of the station meteorological data and GNSS observation data is the same, so there is no need for further unified processing.

[0097] For example, a time window is set for each meteorological profile data, and GNSS observation data and reanalysis meteorological data with sampling points located within the time window are filtered; within the matched time window, meteorological profile data aligned with each sampling point within the time window are generated by interpolation algorithm based on multiple adjacent meteorological profile data.

[0098] This embodiment aligns multi-source heterogeneous data to unify the time resolution and provide data support for subsequent modeling.

[0099] The modeling process of the troposphere will be explained in detail below.

[0100] Based on the composition and spatiotemporal distribution characteristics of the troposphere, the portion of the zenith tropospheric delay that is less affected by water vapor changes due to the dry atmosphere is called the zenith static delay (also known as the zenith dry delay). The portion of the zenith tropospheric delay that is significantly affected by water vapor and is sensitive to water vapor changes is called the zenith wet delay.

[0101] Based on this characteristic, this embodiment will obtain ZHD and ZWD from multiple sources and fuse them into a model to further refine the spatiotemporal variation of zenith tropospheric delay.

[0102] In an optional embodiment, step S103, namely, modeling ZHD and ZWD based on global grid data to obtain the ZHD long-term prediction model and the ZWD short-term prediction model, includes the following steps (1)-(2).

[0103] (1) Based on the reanalysis meteorological data and GNSS observation data in the global grid data, ZHD is modeled to obtain a long-term prediction model for ZHD.

[0104] GNSS observation data and reanalysis meteorological data with a preset time span are labeled according to time points based on periodic changes (such as annual, semi-annual, and seasonal cycles) at network points, and then extracted and fitted to obtain periodic change characteristics. Optionally, the specific implementation process includes the following steps:

[0105] For GNSS ZHD scattered time-series data and reanalysis ZHD grid time-series data with a preset time span, the GNSS ZHD scattered points are interpolated to the corresponding reanalysis grid nodes according to latitude and longitude, forming fused time-series data for each grid point. Using mathematical methods, annual (1-year), semi-annual (6-month), and seasonal periodic signals are separated from each fused time-series data point, and random errors and trend terms are removed. Mathematical methods include Fourier transform, wavelet analysis, and moving average. Each fused time-series data point is then categorized according to the separated periodic signals, resulting in annual, semi-annual, and quarterly groupings. For each group of periodic samples, a mathematical model is established using a fitting function to quantify the periodic variation pattern and obtain the periodic variation characteristics.

[0106] Based on the periodic variation characteristics of ZHD, a long-term prediction model for ZHD is constructed.

[0107] In one optional embodiment, step (1), namely, modeling ZHD based on reanalysis meteorological data and GNSS observation data in global grid data to obtain a long-term prediction model for ZHD, includes the following specific implementation process:

[0108] (1-1) Obtain meteorological data of multiple pressure layers from the reanalysis meteorological data in the global grid data; the meteorological data includes air pressure, potential energy and temperature.

[0109] Meteorological data exists in the form of time-series data.

[0110] (1-2) For each site, the surface layer ZHD of the site is obtained by integrating the meteorological time series data of the multi-pressure layer of the site along the path.

[0111] Optionally, this step is specifically implemented by the following formula:

[0112]

[0113] Among them, 10 -6 The unit conversion factor is used to convert the integration result from the nanometer scale to the meter scale, as the delay of the original calculated value is extremely small. Let be a definite integral function, with the integration interval from the Earth's surface to the top of the atmosphere. The Earth's surface is denoted by s, specifically at altitude z=0 or the station's elevation. The top of the atmosphere is denoted by r, usually taken as the tropopause, approximately 10⁻¹² km. dz is a height infinitesimal element, i.e., an infinitesimal height increment within the integration interval, measured in meters (m). This is an atmospheric physical constant, with a value of approximately 0.776 × 10⁻⁶. -3 hPa -1 It is determined by the refractive index characteristics of dry air. is the gas constant for dry air, with a value of approximately 287 J / (kg·K), and is an inherent physical parameter of dry air. Let be the air pressure at the i-th network point. Let be the temperature of the i-th network point.

[0114] (1-3) Using spatial information as a weighting factor, the ZHD of each station in the GNSS observation data in the global grid data is fused with the surface ZHD of each point in the reanalysis meteorological data through an interpolation algorithm to obtain surface ZHD fused grid data.

[0115] First, determine the latitude and longitude resolution (e.g., 0.25° × 0.25°) and spatial coverage area of ​​the target new grid to define the fusion range. Then, for each node to be analyzed in the target new grid, extract its latitude, longitude, and elevation coordinates. For each node to be analyzed, determine the spatial information of the node and its surrounding GNSS points, and then analyze the spatial information of the grid nodes.

[0116] Spatial information is used to represent the spatial relationship between the node to be determined and the surrounding GNSS scattered points and reanalysis grid nodes.

[0117] For example, spatial information specifically includes planar coordinates and elevation information. Planar coordinates, i.e., latitude and longitude (such as 30°N, 120°E), determine the horizontal position of GNSS scatter points and grid nodes, used to calculate the spatial distance between them. Elevation information, i.e., geodetic elevation (unit: meters), is directly related to atmospheric thickness. Differences in elevation can lead to delays and calculation errors, so a unified benchmark must be established using elevation information to avoid systematic errors.

[0118] For each node to be determined, weights are assigned to the surrounding GNSS scatter points and reanalysis grid nodes based on spatial information (the total weights are 1).

[0119] The weighting factor is a numerical coefficient between 0 and 1 used to measure the degree of influence of a single GNSS scatter point or reanalysis grid node on the ZHD value of the node to be calculated.

[0120] Optionally, the rules for setting weighting factors can be: the closer the distance, the greater the weight; due to the high accuracy of GNSS scatter points, the basic weight ratio can be higher than that of reanalysis meteorological data; the weight of GNSS can be further increased in dense GNSS station areas.

[0121] The interpolation algorithm is invoked to calculate the surface layer fused ZHD value for each node to be determined. Optionally, the interpolation algorithm is inverse distance weighted interpolation or Kriging.

[0122] By repeating this process node by node, the entire target new grid is filled in, thus obtaining the surface layer ZHD fused grid data of the target new grid.

[0123] The surface layer ZHD fused grid data includes the fused ZHD time-series data of each node within a preset time span.

[0124] (1-4) Determine the periodic variation frequency based on the periodic variation characteristics in the ZHD fused grid data of the surface layer.

[0125] For example, the frequency of periodic changes can be annual, semi-annual, or seasonal.

[0126] For example, the periodic frequency can be superimposed using sine or cosine functions to reflect the time variation pattern of ZHD.

[0127] (1-5) Using the Fourier series fitting model, a long-period prediction model for ZHD is constructed based on the ZHD fusion grid data of the surface layer and the periodic change frequency.

[0128] Optionally, the ZHD long-term prediction model can be expressed by the following formula:

[0129]

[0130] in, The value is the predicted zenith static delay at time t for a certain network point, in meters (m). is a constant term, representing the time average of ZHD over many years (baseline value), and is the fundamental component of the model. n is the frequency term number, n=1,2,…,N (n=1 corresponds to the most important cycle, such as the annual cycle; n=2 corresponds to the second most important cycle, such as the semi-annual cycle, and so on). , The Fourier coefficients (parameters to be determined) of the nth frequency term determine the amplitude and phase of that periodic component. , The larger the value, the more significant the influence of the corresponding period; the phase is determined by... and The ratio determines the time of the periodic peak. x is a time variable, which needs to be converted from actual time to a "relative time value" (such as Julian day, annual day, in days; or converted to dimensionless time, such as t / T, where T is the base period). T is the base period, usually taken as 1 year (365 days). cos(nπx / T) and sin(nπx / T) are the cosine and sine basis functions of the nth frequency, respectively capturing the cosine and sine variation trends of the period, and are superimposed to completely replicate the periodic characteristics.

[0131] By using Fourier series decomposition, the periodic changes of ZHD are fitted by superimposing multiple simple harmonic oscillations (sine + cosine terms) of different frequencies, while retaining the constant term (mean) to quantify the time evolution of ZHD.

[0132] (2) Based on the reanalysis meteorological data and meteorological profile data in the global grid data, ZWD is modeled to obtain a short-term ZWD prediction model.

[0133] In an optional embodiment, step (2), namely, modeling ZWD based on reanalysis meteorological data and meteorological profile data in global grid data to obtain a short-term ZWD prediction model, includes the following specific implementation process:

[0134] (2-1) Select multiple pressure layers as standard pressure layers from the reanalysis meteorological data, and resample the data of each pressure layer in the meteorological profile data to the standard pressure layer through the interpolation algorithm to obtain standardized meteorological profile data.

[0135] There is a problem of inconsistency in the pressure layers in the vertical direction between meteorological profile data and reanalysis meteorological data, so it is necessary to standardize the pressure layers of meteorological profile data.

[0136] (2-2) An attention-weighted mechanism is used to weight and fuse the standardized meteorological profile data and the reanalysis meteorological data to obtain the fused meteorological grid data.

[0137] Specifically, the three-dimensional grid of the reanalysis meteorological data is used as the background value, and the standardized meteorological profile data and the reanalysis meteorological data are fused using attention-weighted fusion (Channel Attention) based on a multi-layer neural network.

[0138] Optionally, each point with geographic information can be used as a spatial marker, and radiosonde scatter profiles that were not involved in modeling within the region can be selected as independent spatiotemporal samples to verify the fused meteorological grid data, so as to ensure the quality of the fused data.

[0139] By weighted fusion of data sources and feature extraction, we can achieve effective complementarity and unified modeling of tropospheric information from multiple sources, accurately describe the changes in temperature, humidity and pressure profiles in the vertical direction of the troposphere, and realize high-precision three-dimensional modeling of tropospheric delay in areas with large topographic elevation differences.

[0140] (2-3) For each point in the meteorological grid data, the path integral in the vertical direction is performed based on the meteorological data of the point to obtain the ZWD time series data of the point.

[0141] The delay contribution of water vapor at different altitudes to GNSS signals is accumulated. For each point with geographic information, the corresponding ZWD value is obtained after integration, and finally, the ZWD time series data of the entire region is formed.

[0142] The physical meaning of the integral is that ZWD is the sum of signal delays caused by water vapor. It is necessary to successively accumulate the water vapor delay contribution of each height element (dz) through integration, which is more in line with the actual situation of vertical water vapor distribution than empirical models.

[0143] Alternatively, this step can be implemented using the following formula:

[0144]

[0145] Among them, 10 -6 The unit conversion factor converts the integral result from nanometers to meters to match the actual numerical range of ZWD. The integral function is definite, with the integration interval from the Earth's surface to the top of the atmosphere. The Earth's surface is denoted by 's', specifically at altitude z=0 or the station's elevation. The top of the atmosphere is denoted by 'r', typically taken as the tropopause, approximately 10⁻¹² km. 'dz' represents the infinitesimal height increment within the integration interval, measured in meters (m). K₁, K₂, and K₃ are atmospheric physical constants, with the standard value being: K₁ = 0.776 × 10⁻⁶. - 3 hPa -1 K2 = 0.716 × 10 -3 hPa -1 K3 = 3.75 × 10 5 hPa·K -1 M w M represents the molar mass of water vapor, approximately 18.015 g / mol. d P represents the molar mass of dry air, approximately 28.964 g / mol. w Z represents the vapor pressure of the i-th pressure layer, expressed in hectopascals (hPa), derived from the specific humidity and atmospheric pressure of a three-dimensional grid. w T is the compressibility factor for water vapor, describing the degree to which water vapor deviates from an ideal gas. It is usually approximated as 1 (water vapor in the atmosphere is mostly close to an ideal gas, so the error can be ignored). i The absolute temperature of the i-th pressure layer is expressed in Kelvin (K), and is directly taken from the temperature data of the three-dimensional grid.

[0146] Among them, water vapor pressure P w It can be obtained from the specific humidity, temperature, and air pressure of the corresponding standard atmospheric pressure layer in the meteorological data of the station.

[0147] (2-4) For each grid point in the meteorological grid data, a Gaussian function is used to construct a short-term ZWD prediction model corresponding to the grid point based on the ZWD time series data of the grid point.

[0148] Optionally, a time window can be pre-defined. For "short-term" forecasts (1-24 hours), it's not necessary to use all historical data; instead, select data from a recent, appropriately long time window to train the model. For example, use data from the past 72 hours (3 days) to predict ZWD for the next hour. Select a training set from the ZWD time-series data within the chosen time window. Choose a suitable kernel function. Commonly used kernel functions for time series include the Squared Exponential Kernel and the Matern kernel. Use the training set to train a Gaussian model. The trained model is then the short-term ZWD prediction model.

[0149] Since the starting surface for reanalysis meteorological data is the surface or the elevation corresponding to 1000 hPa, which differs from the actual station elevation, and the water vapor variation near the surface is an important source of error affecting tropospheric delay modeling, this embodiment also constructs a correction model that varies with elevation in the vertical direction.

[0150] In an optional embodiment, step S104, namely the specific implementation process of constructing a correction model for vertical changes with elevation based on global grid data, includes the following steps (1)-(2):

[0151] (1) Data fusion is performed based on global grid data to obtain meteorological grid data, surface ZHD fused grid data and ZWD fused grid data.

[0152] The implementation method for this step is the same as described in the above embodiments, and will not be repeated here.

[0153] (2) Using a neural network algorithm, a correction model for vertical changes with elevation is constructed based on meteorological grid data, surface ZHD fused grid data and ZWD fused grid data.

[0154] The input to a neural network algorithm is a feature vector. The output is the fitted values ​​of key parameters such as specific humidity and temperature. The model automatically extracts surface features (near-surface q, T) and profile features (vertical distribution patterns of q, T, Z of each pressure layer), and learns the correlation between elevation (hgt) and delay to construct a correction model that varies with elevation.

[0155] The neural network algorithm learns the nonlinear relationship between hgt (station potential energy height) and delay, constructs a correction model, and automatically adjusts the delay value by inputting station hgt to offset the error caused by the elevation difference.

[0156] Optionally, the neural network algorithm is a back propagation (BP) neural network model.

[0157] Alternatively, the modified model can be expressed by the following formula:

[0158]

[0159] The geographic information includes longitude (lon), latitude (lat), and potential energy height (hgt); the time information includes annual day (doy) and daily hour (hod); the target variables include specific humidity (ROq), q1…qn, and temperature.

[0160] Based on the model provided in this embodiment, ZHD can be predicted.

[0161] In one optional embodiment, step S105, namely the specific implementation process of ZTD prediction for the site under test based on the ZHD long-term prediction model, the ZWD short-term prediction model, and the modified model, includes:

[0162] Obtain the spatiotemporal information of the site to be tested; the spatiotemporal information is used to characterize the geographical location and the time of the test.

[0163] Spatiotemporal information is input into the ZHD long-term prediction model to obtain the ZHD prediction value of the site to be tested.

[0164] The spatiotemporal information is input into the ZWD short-time prediction model to obtain the ZWD prediction value of the site under test;

[0165] Spatiotemporal information is input into the correction model to obtain correction values;

[0166] Based on the ZHD predicted value, ZWD predicted value, and correction value, determine the ZTD predicted value of the site to be tested.

[0167] Optionally, the spatiotemporal information includes longitude, latitude, and time point.

[0168] By outputting tropospheric static delay and wet delay based on the spatiotemporal information of the station to be measured, and then using a correction model to correct the error caused by the difference between the station height and the modeling reference plane, tropospheric correction information is provided for navigation and positioning in the region.

[0169] It can provide more accurate tropospheric correction values ​​for high-precision navigation and positioning, thereby accelerating the convergence speed of precise positioning. In addition, it can select a global two-dimensional grid with a specific spatial resolution to output tropospheric static delay and wet delay, and visualize them, thus constructing a tropospheric spatiotemporal monitoring platform.

[0170] Figure 2 A schematic diagram of the overall ZTD prediction process provided in this application is shown below. Figure 2 As shown, in this embodiment... Figure 1Based on the implementation examples, the overall process of ZTD prediction is explained. First, multi-source heterogeneous data is preprocessed and aligned. Then, through data fusion and modeling, a long-term ZHD prediction model and a short-term ZWD prediction model are constructed. Furthermore, by constructing a correction model for vertical elevation changes, terrain adaptation is achieved.

[0171] Figure 3 A schematic diagram of the ZTD prediction device provided in this application is shown below. Figure 3 As shown, the ZTD prediction device 30 provided in this embodiment includes:

[0172] Acquisition module 301 is used to acquire multi-source heterogeneous data;

[0173] Alignment module 302 is used to align multi-source heterogeneous data to obtain global grid data;

[0174] Modeling module 303 is used to model the zenith static delay (ZHD) and zenith wet delay (ZWD) based on global grid data to obtain a long-term prediction model for ZHD and a short-term prediction model for ZWD.

[0175] Modeling module 303 is also used to build a corrected model for vertical elevation variations based on global grid data;

[0176] The prediction module 304 is used to perform ZTD prediction for the site under test based on the ZHD long-term prediction model, the ZWD short-term prediction model and the correction model.

[0177] In one possible implementation, the multi-source heterogeneous data includes GNSS observation data from a Global Navigation Satellite System and meteorological data from various stations; the acquisition module 301, after acquiring the multi-source heterogeneous data, is further used for:

[0178] Precise point positioning is performed based on GNSS observation data to obtain the ZTD of each station;

[0179] The ZHD of each station was determined based on the meteorological data from the stations and the Sastamonin model.

[0180] For each site, the difference between the site's ZTD and ZHD is defined as ZWD;

[0181] The ZHD and ZWD of each station were determined as the preprocessed GNSS observation data.

[0182] In one possible implementation, the multi-source heterogeneous data includes meteorological profile data, GNSS observation data, and reanalysis meteorological data;

[0183] Alignment module 302 is used for:

[0184] The GNSS observation data and reanalysis meteorological data are processed with unified time resolution;

[0185] Meteorological profile data is matched with GNSS observation data and reanalysis meteorological data with unified time resolution through time window matching, and then aligned using an interpolation algorithm to obtain global grid data with unified time resolution.

[0186] In one possible implementation, modeling module 303 is used for:

[0187] Based on reanalysis meteorological data and GNSS observation data from global grid data, ZHD is modeled to obtain a long-term prediction model for ZHD.

[0188] Based on reanalysis meteorological data and meteorological profile data from global grid data, ZWD is modeled to obtain a short-term ZWD forecast model.

[0189] In one possible implementation, the modeling module 303, when modeling ZHD based on reanalysis meteorological data and GNSS observation data from global grid data to obtain a long-term ZHD prediction model, is specifically used for:

[0190] Meteorological data from multiple pressure layers at various points are obtained from reanalysis meteorological data in global grid data; the meteorological data includes air pressure, potential energy and temperature.

[0191] For each site, the surface layer ZHD of the site is obtained by integrating the meteorological time series data of the multi-pressure layer of the site along the path.

[0192] Using spatial information as a weighting factor, the ZHD of each station in the GNSS observation data in the global grid data is fused with the surface ZHD of each point in the reanalysis meteorological data through an interpolation algorithm to obtain surface ZHD fused grid data.

[0193] The frequency of periodic changes is determined based on the periodic variation characteristics in the ZHD fused grid data of the surface layer;

[0194] A long-period prediction model for ZHD was constructed by using a Fourier series fitting model based on the ZHD fusion grid data of the surface layer and the periodic variation frequency.

[0195] In one possible implementation, the modeling module 303, when modeling ZWD based on reanalysis meteorological data and meteorological profile data from global grid data to obtain a short-term ZWD forecasting model, is specifically used for:

[0196] Multiple pressure layers are selected as standard pressure layers from the reanalysis meteorological data. The data of each pressure layer in the meteorological profile data are resampled to the standard pressure layer by the interpolation algorithm to obtain standardized meteorological profile data.

[0197] An attention-weighted mechanism is used to weight and fuse standardized meteorological profile data and reanalysis meteorological data to obtain fused meteorological grid data.

[0198] For each point in the meteorological grid data, the path integral in the vertical direction is performed based on the meteorological data of the point to obtain the ZWD time series data of the point;

[0199] For each point in the meteorological grid data, a Gaussian function is used to construct a short-term ZWD prediction model for each point based on the ZWD time series data of the point.

[0200] In one possible implementation, modeling module 303, when constructing a corrected model for vertical elevation variations based on global grid data, is specifically used for:

[0201] Data fusion is performed based on global grid data to obtain meteorological grid data, surface ZHD fused grid data, and ZWD fused grid data;

[0202] Using a neural network algorithm, a correction model for vertical elevation variations is constructed based on meteorological grid data, surface ZHD fused grid data, and ZWD fused grid data.

[0203] In one possible implementation, the prediction module 304 is used for:

[0204] Obtain the spatiotemporal information of the site to be tested; the spatiotemporal information is used to characterize the geographical location and the time of the test.

[0205] Spatiotemporal information is input into the ZHD long-term prediction model to obtain the ZHD prediction value of the site to be tested.

[0206] The spatiotemporal information is input into the ZWD short-time prediction model to obtain the ZWD prediction value of the site under test;

[0207] Spatiotemporal information is input into the correction model to obtain correction values;

[0208] Based on the ZHD predicted value, ZWD predicted value, and correction value, determine the ZTD predicted value of the site to be tested.

[0209] The ZTD prediction device 30 provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0210] Figure 4A schematic diagram of the structure of the electronic device provided in this application. Figure 4 As shown, the electronic device 40 provided in this embodiment includes a processor 401 and a memory 402 that is communicatively connected to the processor 401.

[0211] Optionally, the electronic device 40 also includes a communication component 403. The processor 401, memory 402, and communication component 403 are connected via a bus.

[0212] In the specific implementation process, the processor 401 executes the computer execution instructions stored in the memory 402, causing the processor 401 to perform the above-mentioned method.

[0213] The specific implementation process of processor 401 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0214] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0215] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0216] Buses can be Industry Standard Architecture (ISA) buses, Peripheral Component Interconnect (PCI) buses, or Extended Industry Standard Architecture (EISA) buses, etc. Buses can be categorized into address buses, data buses, control buses, etc.

[0217] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0218] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0219] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as read-only memory (ROM), static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0220] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an application-specific integrated circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0221] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0222] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0223] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0224] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a device to execute all or part of the steps of the methods of the various embodiments of the present invention.

[0225] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for predicting a zenith tropospheric delay (ZTD), characterized in that, The method comprises the following steps: acquiring multi-source heterogeneous data; aligning the multi-source heterogeneous data to obtain global grid data; modeling zenith hydrostatic delay ZHD and zenith wet delay ZWD according to the global grid data to obtain a long-period prediction model of ZHD and a short-time prediction model of ZWD; constructing a correction model varying with height in the vertical direction according to the global grid data; predicting ZTD of a to-be-measured station according to the long-period prediction model of ZHD, the short-time prediction model of ZWD and the correction model.

2. The method of claim 1, wherein, The multi-source heterogeneous data comprises global navigation satellite system (GNSS) observation data and station meteorological data; After acquiring the multi-source heterogeneous data, the method further comprises the following steps: performing precise point positioning according to the GNSS observation data to obtain ZTD of each station; determining ZHD of each station according to the station meteorological data and the Saastamoinen model; determining ZWD of each station by subtracting ZHD from ZTD of the station; determining ZHD and ZWD of each station as preprocessed GNSS observation data.

3. The method of claim 1, wherein, The multi-source heterogeneous data comprises meteorological profile data, GNSS observation data and reanalysis meteorological data; The step of aligning the multi-source heterogeneous data to obtain global grid data comprises the following steps: unifying the time resolution of the GNSS observation data and the reanalysis meteorological data; performing time window matching on the meteorological profile data, the GNSS observation data and the reanalysis meteorological data after unifying the time resolution, and aligning the data by an interpolation algorithm to obtain global grid data with unified time resolution.

4. The method of claim 1, wherein, The step of modeling ZHD and ZWD according to the global grid data to obtain a long-period prediction model of ZHD and a short-time prediction model of ZWD comprises the following steps: modeling ZHD according to the reanalysis meteorological data and the GNSS observation data in the global grid data to obtain the long-period prediction model of ZHD; modeling ZWD according to the reanalysis meteorological data and the meteorological profile data in the global grid data to obtain the short-time prediction model of ZWD.

5. The method of claim 4, wherein, The step of modeling ZHD according to the reanalysis meteorological data and the GNSS observation data in the global grid data to obtain the long-period prediction model of ZHD comprises the following steps: acquiring meteorological data of a plurality of pressure layers of each network point from the reanalysis meteorological data in the global grid data; the meteorological data comprises air pressure, potential energy and temperature; integrating the meteorological time series data of the plurality of pressure layers of each network point along a path to obtain ZHD of a surface layer of the network point; fusing ZHD of each station in the GNSS observation data in the global grid data and ZHD of the surface layer of each network point in the reanalysis meteorological data by an interpolation algorithm to obtain surface layer ZHD fused grid data, with spatial information as a weight factor; determining a periodic change frequency according to the periodic change characteristics of the surface layer ZHD fused grid data; The ZHD long-period prediction model is constructed according to the surface layer ZHD fusion grid data and the periodic change frequency by using a Fourier series fitting model.

6. The method of claim 1, wherein, The ZWD is modeled according to the reanalysis meteorological data and the meteorological profile data in the global grid data to obtain a ZWD short-time prediction model, including: A plurality of pressure layers are selected as standard pressure layers from the reanalysis meteorological data, and the data of each pressure layer in the meteorological profile data is resampled to the standard pressure layers by an interpolation algorithm to obtain standardized meteorological profile data; The standardized meteorological profile data and the reanalysis meteorological data are weighted and fused by using an attention weighting mechanism to obtain fused meteorological grid data; For each grid point in the meteorological grid data, path integration in the vertical direction is performed according to the meteorological data of the grid point to obtain ZWD time series data of the grid point; For each grid point in the meteorological grid data, a Gaussian function is used to construct a ZWD short-time prediction model corresponding to the grid point according to the ZWD time series data of the grid point.

7. The method of claim 1, wherein, The correction model for the vertical direction with elevation change is constructed according to the global grid data, including: Data fusion is performed according to the global grid data to obtain meteorological grid data, surface layer ZHD fusion grid data and ZWD fusion grid data; A neural network algorithm is used to construct a correction model for the vertical direction with elevation change according to the meteorological grid data, surface layer ZHD fusion grid data and ZWD fusion grid data.

8. The method of claim 1, wherein, The ZTD of the to-be-measured station is predicted according to the ZHD long-period prediction model, the ZWD short-time prediction model and the correction model, including: Obtaining the space-time information of the to-be-measured station; the space-time information is used to represent the geographical position and the to-be-measured time; The space-time information is input into the ZHD long-period prediction model to obtain the ZHD prediction value of the to-be-measured station; The space-time information is input into the ZWD short-time prediction model to obtain the ZWD prediction value of the to-be-measured station; The space-time information is input into the correction model to obtain a correction value; The ZTD prediction value of the to-be-measured station is determined according to the ZHD prediction value, the ZWD prediction value and the correction value.

9. A device for predicting a zenith tropospheric delay (ZTD), characterized by, Including: An acquisition module is configured to acquire multi-source heterogeneous data; An alignment module is configured to align the multi-source heterogeneous data to obtain global grid data; A modeling module is configured to model zenith hydrostatic delay (ZHD) and zenith wet delay (ZWD) according to the global grid data to obtain a ZHD long-period prediction model and a ZWD short-time prediction model; The modeling module is further configured to construct a correction model for the vertical direction with elevation change according to the global grid data; A prediction module is configured to predict ZTD of a to-be-measured station according to the ZHD long-period prediction model, the ZWD short-time prediction model and the correction model.

10. An electronic device, comprising: Including: A processor and a memory connected to the processor in communication; The memory stores computer execution instructions; The processor executes computer-executable instructions stored in the memory such that the processor performs the method of any of claims 1-8.

11. A computer readable storage medium, characterized in that, The computer-readable storage medium has stored therein computer-executable instructions that, when executed by a processor, perform the method of any of claims 1-8.

12. A computer program product, characterised in that, A computer program that, when executed by a processor, performs the method of any of claims 1-8.

Images