A real-time ZTD regional modeling method based on GNSS and ground meteorological station observations
By combining GNSS and ground meteorological station observations, a real-time ZTD area modeling method was constructed, which solved the problem of insufficient detailed characteristics of ZTD changes in large height difference areas, improved the real-time estimation accuracy and reliability of ZTD, and achieved high precision and high accuracy Monitoring of spatiotemporal resolution.
Patent Information
- Application Number
- CN202411328326.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-24
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-09-24
AI Technical Summary
In complex environments such as large height difference, traditional tropospheric delay empirical models are difficult to describe the detailed characteristics of ZTD changes, resulting in insufficient accuracy and availability of real-time ZTD models, and lack of real-time ZTD estimation research combining GNSS and ground meteorological station observations. .
Using the real-time ZTD area modeling method based on GNSS and ground meteorological station observation, by calculating the real-time ZTD of each GNSS station and ground meteorological station in the study area, a vertical correction model and a simplified spherical crown harmony analysis model are constructed to unify GNSS and ground meteorological stations. Real-time ZTD is reached to the average elevation datum, and real-time ZTD at the ground meteorological station location is calculated through the weighted average temperature model and the water vapor direct reduction rate correction model.
It improves the accuracy and reliability of ZTD real-time estimation in complex environments such as large height difference, realizes the complementary advantages of GNSS and ground meteorological station observations, and realizes high-precision and high-temporal and spatial resolution real-time monitoring of regional ZTD.
Smart Images

Figure CN118840499B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of GNSS meteorology, and in particular to a real-time ZTD regional modeling method based on GNSS and ground meteorological station observations. Background Art
[0002] Tropospheric delay is one of the important error sources of space geodetic technology. In GNSS navigation and positioning, empirical models are usually used to calculate the initial value of the zenith tropospheric delay (ZTD), and the residual is solved together with other unknown parameters. The ZTD of a GNSS station is generally around 2.4m, consisting of two parts: the zenith hydrostatic delay (ZHD) and the zenith wet delay (ZWD). High-precision tropospheric delay can effectively eliminate the correlation between the tropospheric delay and other parameters, and improve the convergence speed and positioning accuracy of GNSS precise point positioning (PPP). In addition, the tropospheric delay also contains effective information about the neutral atmosphere, which is of great significance for regional water vapor detection and short-term precipitation forecasting. With the development of GNSS and reanalysis data, more and more empirical models of tropospheric delay have been proposed. According to the input of meteorological parameters, the empirical model of tropospheric delay can be divided into a model with measured meteorological parameters and an empirical model without meteorological parameters. By inputting accurate measured meteorological data and location parameters, the accuracy of the measured meteorological parameter model outputting ZTD can reach the centimeter level. The empirical model without meteorological parameters mainly relies on the output values of GNSS ZTD products, sounding data, reanalysis data or numerical weather prediction (NWP) models to model ZTD in the spatial domain and time domain respectively. In addition, the empirical values of meteorological parameters provided by the meteorological parameter model can also be used to calculate ZTD using the measured meteorological parameter model.
[0003] For real-time tropospheric delay applications, due to the limitations of data sources and construction strategies, traditional empirical models of tropospheric delay often have difficulty describing the detailed characteristics of ZTD changes. In order to ensure the accuracy and reliability of real-time ZTD services, regional real-time ZTD correction models are usually constructed based on GNSS Continuous Operational Reference System (CORS) observation data. In addition, NWP forecast products have been effectively used in real-time tropospheric delay correction algorithms due to their good spatial coverage and service continuity. However, in areas with large elevation difference terrain or missing field data, the accuracy of NWP forecast products may be affected by deviations. If only GNSS observations are used for real-time tropospheric delay applications, the constructed real-time ZTD model will also be insufficient in accuracy and availability due to problems such as large elevation differences and uneven distribution of GNSS stations. In fact, ground meteorological stations are another widely distributed site observation system besides GNSS, which can provide real-time meteorological observations with high accuracy and high temporal resolution at the site location. Through the measured meteorological parameter model, the real-time ZTD at the location of the ground meteorological station can be calculated using temperature, air pressure and relative humidity. Unfortunately, there is currently no research on combining GNSS and ground-based meteorological station observations for real-time ZTD estimation, resulting in the unclear enhancement effect of ground-based meteorological station observations on regional real-time ZTD estimation.
[0004] In view of the importance of regional real-time ZTD estimation research and the obvious changes in ZTD in complex environmental areas such as large elevation differences, how to estimate the real-time ZTD at the location of ground meteorological stations and combine GNSS and ground meteorological station ZTD for modeling are urgent problems that need to be solved in regional ZTD high-precision and high spatiotemporal resolution monitoring. It has important theoretical research and practical application significance for promoting the performance improvement of GNSS real-time applications and the effective integration of different data source products. Summary of the invention
[0005] In order to effectively improve the accuracy and reliability of real-time ZTD estimation in complex environmental areas such as large elevation differences, the present invention provides a real-time ZTD regional modeling method based on GNSS and ground meteorological station observations, which solves the problem that a single data source is difficult to describe the detailed characteristics of real-time changes of ZTD in complex environmental areas such as large elevation differences due to limitations in site distribution and construction strategies.
[0006] To achieve the above object, the present invention provides the following technical solution: a real-time ZTD regional modeling method based on GNSS and ground meteorological station observations, comprising the following steps:
[0007] S1. Calculate the real-time ZTD at each GNSS station and ground meteorological station in the study area, where ZTD represents the zenith tropospheric delay;
[0008] S2, construct a vertical correction model for ZTD in the study area, and unify the real-time ZTD of GNSS and ground meteorological stations to the mean elevation datum;
[0009] S3. Construct a simplified spherical cap harmonic analysis model, fit the real-time ZTD of GNSS and ground meteorological stations on the mean elevation datum, solve the coefficients of the simplified spherical cap harmonic analysis model, and obtain the real-time ZTD model of the study area.
[0010] Preferably, the specific steps of step S1 are as follows:
[0011] S11. Calculate the real-time ZTD values at all GNSS stations in the study area through real-time precise point positioning y G,s ;
[0012] S12. Construct a weighted average temperature model for the study area;
[0013] S13. Construct a water vapor lapse rate correction model for the study area;
[0014] S14. Calculate the real-time ZTD values at all ground meteorological stations in the study area y M,s .
[0015] Preferably, in step S12, constructing a weighted average temperature model of the study area specifically includes the following steps:
[0016] Constructing the weighted average temperature of the station based on the sounding data of the sounding stations in the study area T m With site temperature T s The linear regression model between T m Calculate using hierarchical integration:
[0017] (1)
[0018] In formula (1), dh is the height difference between the pressure layers in the sounding data, e s,k and T s,k are the water vapor pressure and temperature of each atmospheric pressure layer of the sounding data, respectively. The sum of the water vapor pressure or temperature of the upper and lower layers is divided by 2, which is the mean. is the total number of pressure stratification layers of the sounding data;
[0019] Vapor Pressure e s,k The calculation formula is as follows:
[0020] (2)
[0021] In formula (2), RH s,k is the relative humidity of the pressure stratification corresponding to the sounding data;
[0022] According to the site temperature T s and the calculated T m Time series, using the least squares method to estimate parameters, obtain the study area T m -T s The linear regression model is as follows:
[0023] (3)
[0024] In formula (3), a and b are the estimated linear regression model coefficients.
[0025] Preferably, in step S13, constructing a water vapor lapse rate correction model for the study area specifically includes the following steps:
[0026] The output value of the GPT3 model is used as the initial value, and the water vapor lapse rate correction model of the study area is constructed using the reanalysis data. The construction process of the correction model is as follows:
[0027] Use DEM products to obtain the ground height at the grid points of the reanalysis data in the study area h s , interpolate the air pressure, temperature and relative humidity at the ground height of the grid points based on the reanalysis profile data to form a virtual pressure layer;
[0028] The Saastamoinen model is used to calculate the ZWD above the top layer of the reanalysis data. d , where ZWD is the zenith wet delay:
[0029] (4)
[0030] In formula (4), φ and h s are latitude and geodetic height respectively, T c and e s,t are the temperature and water vapor pressure at the top layer, respectively;
[0031] The ZWD from the top floor to the station height is calculated using the layered integration method. f :
[0032] (5)
[0033] Among them, the calculation formula of the layered integration method ZHD, where ZHD represents the zenith dry delay:
[0034] (6)
[0035] In formulas (5) and (6), k 1=77.604 K / hPa, k 2=64.79 K / hPa, k 3=377600.0 k 2 / hPa, e s,i and T s,i are the water vapor pressure and temperature of each reanalysis pressure layer, P s,i is the pressure of each reanalysis pressure layer, m is the number of pressure stratification layers from the top layer of reanalysis data to the station height;
[0036] ZWD d and ZWD f The ZWD time series at each grid point is obtained by adding them together; the weighted average temperature at the ground location of the reanalysis data is calculated based on formulas (1) and (2): T m,z as well as e s Time series, using Askne-Nordius model to calculate the water vapor loss rate of reanalysis data λ For time series, the Askne-Nordius model ZWD calculation formula is:
[0037] (7)
[0038] In formula (7), is the empirical coefficient of atmospheric refractive index, R d is the gas constant for dry air, R v is the gas constant of water vapor; R = 8.31434 J / (mol·K), which is the universal constant of an ideal gas; M d is the molar mass of dry air, equal to 28.9644×10 -3 kg / mol; g m is the gravitational acceleration at the center of mass of the vertical atmospheric column, about 9.80065 m / s 2 ;
[0039] Water vapor lapse rate λ Time series minus the GPT3 model λ Time series, calculate the Δ for each grid point λ Time series, and model it using a harmonic function containing a mean term, an annual cycle term, and a semi-annual cycle term, as follows:
[0040] (8)
[0041] In the formula, c 0. c 1. c 2. c 3. c 4 are model coefficients, t is the time variable, the unit is day; ε Δλ is the model residual.
[0042] Preferably, in step S14, the real-time ZTD values at all ground meteorological stations in the study area are calculated. y M,s , specifically including the following steps:
[0043] According to the real-time air pressure P provided by the ground weather station s,M ,latitude φ M and the earth high h M , use the Saastamoinen model ZHD calculation formula to calculate the real-time ZHD of the ground weather station M :
[0044] (9)
[0045] For ZWD M The calculation of the water vapor lapse rate of the GPT3 model at the ground meteorological station is first calculated. Then, the correction values of the four nearest grid points are calculated based on the water vapor lapse rate correction model. The correction value at the ground meteorological station is calculated using the bilinear interpolation method. The correction value is added to the water vapor lapse rate of the GPT3 model to obtain the water vapor lapse rate at the ground meteorological station. Then, the water vapor pressure and weighted average temperature at the ground meteorological station are calculated according to formulas (2) and (3). Finally, the ZWD at the ground meteorological station is calculated according to formula (7). M , ZHD M and ZWD M Add together to get the real-time ZTD value at all ground weather stations y M,s .
[0046] Preferably, the specific steps of step S2 are as follows:
[0047] S21. Calculate the ZTD at the pressure layer 8 km and below for all ERA5 grid points in the study area;
[0048] S22, constructing a ZTD vertical correction function;
[0049] S23, constructing a seasonal model of the ZTD vertical correction function coefficients, and calculating the seasonal model coefficients;
[0050] S24. Unify the real-time ZTD of GNSS and ground meteorological stations to the mean elevation datum.
[0051] Preferably, in step S22, a ZTD vertical correction function is constructed as follows:
[0052] (10)
[0053] In the formula, z 0 and z 3 are the ZTD values at altitudes of 0 km and 3 km respectively, β 1. β 2 and β 3 are the vertical variation function coefficients.
[0054] Preferably, in step S23, constructing a seasonal model of the ZTD vertical correction function coefficients and calculating the seasonal model coefficients specifically includes the following steps:
[0055] Based on the reanalysis data, the atmospheric pressure layer data is calculated based on the grid points in the study area. β 1. β 2 and β 3, using the harmonic function containing the mean term, the annual cycle term and the semi-annual cycle term to model them respectively, as follows:
[0056] (11)
[0057] In the formula, i = 1, 2, 3, a i,1 , b i,3 , c i,3 , b i,4 , c i,4 All are model coefficients, and the least square method is used for parameter estimation to calculate seasonal model coefficients; ε β The residuals from the model fit.
[0058] Preferably, in step S24, the real-time ZTD of the GNSS and the ground meteorological station is unified to the average elevation reference plane, which specifically includes the following steps:
[0059] Calculate the average of the GNSS and ground weather station elevations as the elevation datum, based on the real-time ZTD value at the GNSS position y G,s and the real-time ZTD value at the ground weather station location y M,s , use formulas (10) and (11) to calculate the GNSS real-time ZTD values of the nearest four grid points on the elevation reference surface y G and real-time ZTD value of ground weather station y M .
[0060] Preferably, the specific steps of step S3 are as follows:
[0061] S31. Construct a simplified spherical cap harmonic analysis model:
[0062] The geodetic coordinates of each GNSS and ground meteorological station in the study area ( B , L ) into geocentric coordinates ( θ , λ ):
[0063] (12)
[0064] In formula (12), B , L They are the latitude and longitude in the WGS84 coordinate system respectively; θ, λ are the latitude and longitude in the geocentric coordinate system respectively; e is the first eccentricity of the WGS84 ellipsoid, e 2 = 0.00669438002;
[0065] Secondly, according to the geocentric coordinates of the spherical cap pole ( θ p , λ p ) Calculate the spherical crown coordinates of each station ( θ c , λ c ):
[0066] (13)
[0067] In formula (13), θ c and λc The value ranges are [0~ θ 0] and [0~2π], θ 0 is the radius of the spherical cap;
[0068] Again, map the spherical cap co-latitude to [0~π], and the converted co-latitude θ w for:
[0069] (14)
[0070] Finally, ordinary Legendre functions are used instead of non-integer-order Legendre functions to construct a simplified spherical cap harmonic analysis model:
[0071] (15)
[0072] (16)
[0073] In formulas (15) and (16), f ( x ) is the observed value, which is the real-time ZTD in the present invention; Y lk ( x ) is based on θ w and λ c Calculated coefficients; c lk are unknown parameters of the model; n is the order; l is the number of times, ranging from [0~ n ], k is an auxiliary value, ranging from [- l ~ l ];| k | indicates k The absolute value of is the normalized Legendre function; is also a function, which means that when k <0, give k Add absolute values to ensure input parameters k is a positive value;
[0074] According to formulas (15) and (16), the observation equation is:
[0075] (17)
[0076] In formula (17), y G and y MThey are the real-time ZTD of GNSS and ground meteorological station after elevation correction, F G and F M is the coefficient vector, α is the parameter vector, ε G and ε M is the residual vector, where the Tikhonov regularization term is introduced to prevent the model from overfitting, that is, the zero observation of the parameter vector is introduced:
[0077] (18)
[0078] In formula (18), μ is the zero vector, ε μ is the residual vector, on this basis, the parameter vector α The least squares numerical solution of is:
[0079] (19)
[0080] In formula (19), λ μ = 0.001 is the empirical Tikhonov regularization parameter, I μ is the unit matrix;
[0081] S32. Calculate the simplified spherical cap harmonic analysis model coefficients based on the real-time ZTD of GNSS and ground meteorological stations on the average elevation surface, and obtain the real-time ZTD model of the study area.
[0082] Compared with the prior art, the present invention has the following beneficial effects:
[0083] 1. The present invention uses the real-time PPP algorithm to calculate the real-time GNSS ZTD, and constructs a regional weighted average temperature model and a water vapor lapse rate correction model. The real-time ZTD at the location of the ground meteorological station is calculated by the Saastamoinen and Askne-Nordius models. Then, a ZTD vertical correction model is established to unify the GNSS ZTD calculated by real-time PPP and the real-time ZTD of the ground meteorological station on the average elevation surface. Finally, the simplified spherical cap harmonic analysis model is used for regional modeling to achieve accurate estimation of the real-time ZTD.
[0084] 2. In the process of constructing the regional real-time ZTD model, the present invention effectively improves the estimation accuracy of the real-time ZTD in the area where GNSS observations are missing by introducing ground meteorological station observations, and realizes the complementary advantages of the two. This is of great significance for realizing high-precision, high-temporal and high-spatial resolution real-time monitoring of regional ZTD. BRIEF DESCRIPTION OF THE DRAWINGS
[0085] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.
[0086] In the attached picture:
[0087] Figure 1 It is a schematic diagram of the flow chart of the modeling method of the present invention. DETAILED DESCRIPTION
[0088] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.
[0089] Example: Figure 1 As shown, a real-time ZTD regional modeling method based on GNSS and ground meteorological station observations includes the following steps:
[0090] S1. Calculate the real-time ZTD at each GNSS station and ground meteorological station in the study area;
[0091] Furthermore, step S1 is specifically as follows:
[0092] S11. Based on the GNSS real-time data stream (such as GPS / GLONASS real-time data stream), broadcast ephemeris, and SSR (State Space Representation) corrections (such as SSR corrections provided by CNES), the real-time precise point positioning (PPP) solution strategy is used to estimate the ZTD parameters and obtain the real-time ZTD values at all GNSS positions. y G,s , where ZTD represents the zenith tropospheric delay;
[0093] S12. Construct a weighted average temperature model for the study area, including:
[0094] Constructing the weighted average temperature of the station based on the sounding data of the sounding stations in the study area T m With site temperature T s The linear regression model between T m Calculate using hierarchical integration:
[0095] (1)
[0096] In formula (1), dh is the height difference between the pressure layers of the sounding data,e s,k and T s,k are the water vapor pressure and temperature of each pressure layer of the sounding data, respectively. The sum of the water vapor pressure or temperature of the upper and lower layers is divided by 2, which is the mean. m t is the total number of pressure stratification layers of the sounding data;
[0097] Vapor Pressure e s,k The calculation formula is as follows:
[0098] (2)
[0099] In formula (2), RH s,k is the relative humidity of the pressure stratification corresponding to the sounding data;
[0100] According to the site temperature T s and the calculated T m Time series, using the least squares method to estimate parameters, the study area T m -T s The linear regression model is as follows:
[0101] (3)
[0102] In formula (3), a and b is the estimated linear regression model coefficient;
[0103] S13. Construct a water vapor lapse rate correction model for the study area, including:
[0104] The output value of the GPT3 model is used as the initial value, and the water vapor lapse rate correction model of the study area is constructed using the reanalysis data. The GPT3 model is a well-known empirical tropospheric delay meteorological parameter model, which is the same as its predecessor, the GPT2w model. Both can provide important parameters in the tropospheric delay calculation process. The construction process of the correction model is as follows:
[0105] Use DEM products to obtain the ground height at the grid points of the reanalysis data in the study area h s , interpolate the air pressure, temperature and relative humidity at the ground height of the grid points based on the reanalysis profile data to form a virtual pressure layer;
[0106] The Saastamoinen model is used to calculate the ZWD above the top layer of the reanalysis data. d , where ZWD is the zenith wet delay:
[0107] (4)
[0108] In formula (4), φ and h s are latitude and geodetic height respectively, T c and e s,t are the temperature and water vapor pressure at the top layer, respectively;
[0109] The ZWD from the top floor to the station height is calculated using the layered integration method. f :
[0110] (5)
[0111] Among them, the calculation formula of the layered integration method ZHD, where ZHD represents the zenith dry delay:
[0112] (6)
[0113] In formula (5), k 1=77.604 K / hPa, k 2=64.79 K / hPa, k 3=377600.0 k 2 / hPa, e s,i and T s,i are the water vapor pressure and temperature of each reanalysis pressure layer, P s,i is the pressure of each reanalysis pressure layer, m is the number of pressure stratification layers from the top layer of reanalysis data to the station height;
[0114] ZWD d and ZWD f The ZWD time series at each grid point is obtained by adding them together; the weighted average temperature at the ground height of the reanalysis grid point is calculated based on formulas (1) and (2): T m,z as well as e s Time series, using Askne-Nordius model to calculate the water vapor loss rate of reanalysis data λ For time series, the Askne-Nordius model ZWD calculation formula is:
[0115] (7)
[0116] In formula (7), is the empirical coefficient of atmospheric refractive index, R d is the gas constant for dry air, R v is the gas constant of water vapor; R = 8.31434 J / (mol·K), which is the universal constant of an ideal gas; M d is the molar mass of dry air, equal to 28.9644×10 -3 kg / mol; g m is the gravitational acceleration at the center of mass of the vertical atmospheric column, about 9.80065 m / s 2 .
[0117] Water vapor lapse rate λ Time series minus the GPT3 model λ Time series, calculate the Δ for each grid point λ time series and model it using a harmonic function that includes a mean term, an annual periodic term, and a semiannual periodic term as follows:
[0118] (8)
[0119] In the formula, c 0. c 1. c 2. c 3. c 4 are model coefficients, t is the time variable, the unit is day; ε Δλ is the model residual.
[0120] S14. Calculate the real-time ZTD values at all ground meteorological stations in the study area y M,s , including:
[0121] According to the real-time air pressure provided by the ground weather station P s,M ,latitude φ M as high as the earth h M , use the Saastamoinen model ZHD calculation formula to calculate the real-time ZHD of the ground weather station M :
[0122] (9)
[0123] For ZWD MThe calculation of the water vapor lapse rate of the GPT3 model at the ground meteorological station is first calculated. Then, the correction values of the four nearest grid points are calculated based on the water vapor lapse rate correction model. The correction value at the ground meteorological station is calculated using the bilinear interpolation method. The correction value is added to the water vapor lapse rate of the GPT3 model to obtain the water vapor lapse rate at the ground meteorological station. Then, the water vapor pressure and weighted average temperature at the ground meteorological station are calculated according to formulas (2) and (3). Finally, the ZWD at the ground meteorological station is calculated according to formula (7). M , ZHD M and ZWD M Add together to get the real-time ZTD value at all ground weather stations y M,s .
[0124] S2, construct a vertical correction model for ZTD in the study area and unify the real-time ZTD of GNSS and ground meteorological stations to the mean elevation datum;
[0125] Furthermore, step S2 is specifically as follows:
[0126] S21. Calculate the ZTD at the pressure layer 8 km and below for all ERA5 grid points in the study area. Specifically, select the pressure layer 8 km and below for all grid points in the study area reanalysis data, and calculate the ZTD at the pressure layer position according to formulas (5) and (6);
[0127] S22. Construct ZTD vertical correction function:
[0128] (10)
[0129] In the formula, z 0 and z 3 are the ZTD values at altitudes of 0 km and 3 km respectively, β 1. β 2 and β 3 are the vertical variation function coefficients;
[0130] S23, constructing a seasonal model of the ZTD vertical correction function coefficients, and calculating the seasonal model coefficients, specifically including:
[0131] Based on the reanalysis data, the atmospheric pressure layer data is calculated based on the grid points in the study area. β 1. β 2 and β The time series of 3 is modeled using harmonic functions containing mean terms, annual cycle terms, and semi-annual cycle terms, as shown below:
[0132] (11)
[0133] In the formula,i = 1, 2, 3, a i,1 , b i,3 , c i,3 , b i,4 , c i,4 All are model coefficients, and the least square method is used for parameter estimation to calculate seasonal model coefficients; ε β are the residuals from the model fit;
[0134] S24. Unify the real-time ZTD of GNSS and ground meteorological stations to the mean elevation datum, including:
[0135] Calculate the average of the GNSS and ground weather station elevations as the elevation datum, based on the real-time ZTD value at the GNSS position y G,s and the real-time ZTD value at the ground weather station location y M,s , use formulas (10) and (11) to calculate the GNSS real-time ZTD values of the nearest four grid points on the elevation reference surface y G and real-time ZTD value of ground weather station y M .
[0136] S3, construct a simplified spherical cap harmonic analysis model, fit the real-time ZTD of GNSS and ground meteorological stations on the average elevation datum, solve the model coefficients, and obtain the real-time ZTD model of the study area;
[0137] Furthermore, step S3 is specifically as follows:
[0138] S31. Construct a simplified spherical cap harmonic analysis model, specifically including:
[0139] The geodetic coordinates of each GNSS and ground meteorological station in the study area ( B , L ) into geocentric coordinates ( θ , λ ):
[0140] (12)
[0141] In formula (12), B , L They are the latitude and longitude in the WGS84 coordinate system respectively; θ, λ are the latitude and longitude in the geocentric coordinate system respectively; e is the first eccentricity of the WGS84 ellipsoid,e 2 = 0.00669438002.
[0142] Secondly, according to the geocentric coordinates of the spherical cap pole ( θ p , λ p ) Calculate the spherical crown coordinates of each station ( θ c , λ c ):
[0143] (13)
[0144] In formula (13), θ c and λ c The value ranges are [0~ θ 0] and [0~2π], θ 0 is the radius of the spherical cap;
[0145] Again, map the spherical cap co-latitude to [0~π], and the converted co-latitude θ w for:
[0146] (14)
[0147] Finally, ordinary Legendre functions are used instead of non-integer-order Legendre functions to construct a simplified spherical cap harmonic analysis model:
[0148] (15)
[0149] (16)
[0150] In formulas (15) and (16), f ( x ) is the observed value, which is the real-time ZTD in the present invention; Y lk ( x ) is based on θ w and λ c Calculated coefficients; c lk are unknown parameters of the model; n is the order; l is the number of times, ranging from [0~ n ], k is an auxiliary value, ranging from [- l ~ l ];| k| indicates k The absolute value of is the normalized Legendre function; is also a function, which means that when k <0, give k Add absolute values to ensure input parameters k is a positive value;
[0151] According to formulas (15) and (16), the observation equation is:
[0152] (17)
[0153] In formula (17), y G and y M They are the real-time ZTD of GNSS and ground meteorological station after elevation correction, F G and F M is the coefficient vector, α is the parameter vector, ε G and ε M is the residual vector, where the Tikhonov regularization term is introduced to prevent the model from overfitting, that is, the zero observation of the parameter vector is introduced:
[0154] (18)
[0155] In formula (18), μ is the zero vector, ε μ is the residual vector, on this basis, the parameter vector α The least squares numerical solution of is:
[0156] (19)
[0157] In formula (19), λ μ = 0.001 is the empirical Tikhonov regularization parameter, I μ is the unit matrix;
[0158] S32. Calculate the simplified spherical cap harmonic analysis model coefficients based on the real-time ZTD of GNSS and ground meteorological stations on the average elevation surface, and obtain the real-time ZTD model of the study area.
[0159] The example verification results of the present invention are as follows:
[0160] In order to verify the accuracy and reliability of the real-time ZTD obtained by the method proposed in the present invention, the present invention uses the hourly value data matching the GNSS and ground meteorological stations from September 17 to December 31, 2023 to construct a real-time ZTD model in Yunnan area for experimental verification.
[0161] The data used include GNSS data, ground meteorological data, sounding data, ERA5 pressure layer data and TanDEM-X 90m DEM products, and the ERA5 ground ZTD is used as the reference truth value. Four schemes are used for comparative analysis: ① GPT3 ZTD; ② Single ground meteorological station model ZTD; ③ Single GNSS station model ZTD; ④ GNSS + ground meteorological station model ZTD. The accuracy evaluation criteria include mean deviation and RMS. The experimental platform of the present invention is Python3.8 software under the Windows operating system, and the analysis results of the present invention are not affected by the operating system and Python software version.
[0162] The experimental results show that the average deviations of the four schemes are 20.95, 3.39, 18.77 and 11.71 mm, respectively, and the average RMS are 35.66, 28.45, 24.82 and 21.20 mm, respectively. According to the calculation results, the GPT3 model has the worst overall accuracy among the four schemes, while the GNSS+ground meteorological station model has the highest accuracy. Compared with the GPT3 model, the single ground meteorological station model and the single GNSS model, the overall accuracy of the ZTD calculated after combining GNSS and ground meteorological station observations has improved by about 40.55%, 25.48% and 14.58%, respectively. This shows that the introduction of ground meteorological station observations has effectively improved the overall accuracy of real-time ZTD estimation in Yunnan.
[0163] In summary, the real-time ZTD regional modeling method based on GNSS and ground meteorological station observations has good performance. The introduction of ground meteorological station observations effectively improves the estimation accuracy of real-time ZTD in areas where GNSS observations are missing, and realizes the complementary advantages of the two. This is of great significance for achieving high-precision, high-temporal and high-spatial resolution real-time monitoring of regional ZTD.
[0164] Finally, it should be noted that the above description is only a preferred example of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A real-time ZTD regional modeling method based on GNSS and ground meteorological station observations, characterized in that: The following steps are involved: S1. Calculate the real-time ZTD values at the locations of each GNSS station and ground meteorological station in the study area, where ZTD represents the zenith tropospheric delay; S2, construct a vertical correction model for ZTD in the study area and unify the real-time ZTD of GNSS and ground meteorological stations to the mean elevation datum; S3. Construct a simplified spherical cap harmonic analysis model, fit the real-time ZTD of GNSS and ground meteorological stations on the average elevation datum, solve the coefficients of the simplified spherical cap harmonic analysis model, and obtain the real-time ZTD model of the study area; The specific steps of step S1 are as follows: S11. Calculate the real-time ZTD values at all GNSS stations in the study area through real-time precise point positioning y G,s ; S12. Construct a weighted average temperature model for the study area; In step S12, a weighted average temperature model of the study area is constructed, which specifically includes the following steps: Constructing the weighted average temperature of the station based on the sounding data of the sounding stations in the study area T m With site temperature T s The linear regression model between T m Calculate using hierarchical integration: (1) In formula (1), dh is the height difference between the pressure layers of the sounding data, e s,k and T s,k are the water vapor pressure and temperature of each pressure layer of the sounding data, respectively. The sum of the water vapor pressure or temperature of the upper and lower layers is divided by 2, which is the mean. m t is the total number of pressure stratification layers of the sounding data; Vapor Pressure e s,k The calculation formula is as follows: (2) In formula (2), RH s,k is the relative humidity of the pressure stratification corresponding to the sounding data; According to the site temperature T s and the calculated T m Time series, using the least squares method to estimate parameters, obtain the study area T m -T s The linear regression model is as follows: (3) In formula (3), a and b is the estimated linear regression model coefficient; S13. Construct a water vapor lapse rate correction model for the study area; In step S13, a water vapor lapse rate correction model for the study area is constructed, which specifically includes the following steps: The output value of the GPT3 model is used as the initial value, and the water vapor lapse rate correction model of the study area is constructed using the reanalysis data. The construction process of the correction model is as follows: Use DEM products to obtain the ground height at the grid points of the reanalysis data in the study area h s , interpolate the air pressure, temperature and relative humidity at the ground height of the grid points based on the reanalysis profile data to form a virtual pressure layer; The Saastamoinen model is used to calculate the ZWD above the top layer of the reanalysis data. d , where ZWD is the zenith wet delay: (4) In formula (4), φ and h s are latitude and geodetic height respectively, T c and e s,t are the temperature and water vapor pressure at the top layer, respectively; The ZWD from the top floor to the station height is calculated using the layered integration method. f, : (5) Among them, the calculation formula of the layered integration method ZHD, where ZHD represents the zenith dry delay: (6) In formulas (5) and (6), k 1=77.604 K / hPa, k 2=64.79 K / hPa, k 3=377600.0 k 2 / hPa, e s,i and T s,i are the water vapor pressure and temperature of each reanalysis pressure layer, P s,i is the pressure of each reanalysis pressure layer, m is the number of pressure stratification layers from the top layer of reanalysis data to the station height; ZWD d and ZWD f The ZWD time series at each grid point is obtained by adding them together; the weighted average temperature at the ground location of the reanalysis data is calculated based on formulas (1) and (2): T m,z as well as e s Time series, using Askne-Nordius model to calculate the water vapor loss rate of reanalysis data λ For time series, the Askne-Nordius model ZWD calculation formula is: (7) In formula (7), is the empirical coefficient of atmospheric refractive index, R d is the gas constant for dry air, R v is the gas constant of water vapor; R =8.31434 J / (mol·K), which is the universal constant of ideal gas; M d is the molar mass of dry air, equal to 28.9644×10 -3 kg / mol; g m is the acceleration due to gravity at the center of mass of the vertical atmospheric column; Water vapor lapse rate λ Time series minus the GPT3 model λ Time series, calculate the Δ for each grid point λ Time series, and model it using a harmonic function containing a mean term, an annual cycle term, and a semi-annual cycle term, as follows: (8) In the formula, c 0. c 1. c 2. c 3. c 4 are model coefficients, t is the time variable, the unit is day; ε Δλ is the model residual; S14. Calculate the real-time ZTD values at all ground meteorological stations in the study area y M,s ; In step S14, the real-time ZTD values at all ground meteorological stations in the study area are calculated. y M,s , specifically including the following steps: According to the real-time air pressure P provided by the ground weather station s,M ,latitude φ M and the earth high h M , use the Saastamoinen model ZHD calculation formula to calculate the real-time ZHD of the ground weather station M : (9) For ZWD M The calculation of the water vapor lapse rate of the GPT3 model at the ground meteorological station is first calculated. Then, the correction values of the four nearest grid points are calculated based on the water vapor lapse rate correction model. The correction value at the ground meteorological station is calculated using the bilinear interpolation method. The correction value is added to the water vapor lapse rate of the GPT3 model to obtain the water vapor lapse rate at the ground meteorological station. Then, the water vapor pressure and weighted average temperature at the ground meteorological station are calculated according to formulas (2) and (3). Finally, the ZWD at the ground meteorological station is calculated according to formula (7). M , ZHD M and ZWD M Add together to get the real-time ZTD value at all ground weather stations y M,s ; The specific steps of step S2 are as follows: S21, calculate the ZTD at the pressure layer of 8 km and below for all ERA5 grid points in the study area; S22, constructing a ZTD vertical correction function; In step S22, a ZTD vertical correction function is constructed as follows: (10) In the formula, z 0 and z 3 are the ZTD values at altitudes of 0 km and 3 km respectively, β 1. β 2 and β 3 are the vertical variation function coefficients; S23, constructing a seasonal model of the ZTD vertical correction function coefficients, and calculating the seasonal model coefficients; In step S23, a seasonal model of the ZTD vertical correction function coefficient is constructed and the seasonal model coefficient is calculated, which specifically includes the following steps: Based on the reanalysis data, the atmospheric pressure layer data is calculated based on the grid points in the study area. β 1. β 2 and β 3, using the harmonic function containing the mean term, the annual cycle term and the semi-annual cycle term to model them respectively, as follows: (11) In the formula, i = 1, 2, 3, a i,1 , b i,3 , c i,3 , b i,4 , c i,4 All are model coefficients, and the least square method is used for parameter estimation to calculate seasonal model coefficients; ε β are the residuals from the model fit; S24. Unify the real-time ZTD of GNSS and ground meteorological stations to the mean elevation datum; In step S24, the real-time ZTD of the GNSS and ground meteorological station is unified to the average elevation reference surface, which specifically includes the following steps: Calculate the average of the GNSS and ground weather station elevations as the elevation datum, based on the real-time ZTD value at the GNSS position y G,s and the real-time ZTD value at the ground weather station location y M,s , use formulas (10) and (11) to calculate the GNSS real-time ZTD values of the nearest four grid points on the elevation reference surface y G and real-time ZTD value of ground weather station y M .
2. The real-time ZTD regional modeling method based on GNSS and ground meteorological station observations according to claim 1, characterized in that: The specific steps of step S3 are as follows: S31. Construct a simplified spherical cap harmonic analysis model: The geodetic coordinates of each GNSS and ground meteorological station in the study area ( B , L ) into geocentric coordinates ( θ , λ ): (12) In formula (12), B , L They are the latitude and longitude in the WGS84 coordinate system respectively; θ、λ are the latitude and longitude in the geocentric coordinate system respectively; e is the first eccentricity of the WGS84 ellipsoid, e 2 = 0.00669438002; Secondly, according to the geocentric coordinates of the spherical cap pole ( θ p , λ p ) Calculate the spherical crown coordinates of each station ( θ c , λ c ): (13) In formula (13), θ c and λ c The value ranges are [0~ θ 0] and [0~2π], θ 0 is the radius of the spherical cap; Again, map the spherical cap co-latitude to [0~π], and the converted co-latitude θ w for: (14) Finally, ordinary Legendre functions are used instead of non-integer-order Legendre functions to construct a simplified spherical cap harmonic analysis model: (15) (16) In formulas (15) and (16), f ( x ) is the observed value, which is the real-time ZTD in the present invention; Y lk ( x ) is based on θ w and λ c Calculated coefficients; c lk are unknown parameters of the model; n is the order; l is the number of times, ranging from [0~ n ], k is an auxiliary value, ranging from [- l ~ l ];| k | indicates k The absolute value of is the normalized Legendre function; is also a function, which means that when k <0, give k Add absolute values to ensure input parameters k is a positive value; According to formulas (15) and (16), the observation equation is: (17) In formula (17), y G and y M They are the real-time ZTD of GNSS and ground meteorological station after elevation correction, F G and F M is the coefficient vector, α is the parameter vector, ε G and ε M is the residual vector, where the Tikhonov regularization term is introduced to prevent the model from overfitting, that is, the zero observation of the parameter vector is introduced: (18) In formula (18), μ is the zero vector, ε μ is the residual vector, on this basis, the parameter vector α The least squares numerical solution of is: (19) In formula (19), λ μ = 0.001 is the empirical Tikhonov regularization parameter, I μ is the unit matrix; S32. Calculate the simplified spherical cap harmonic analysis model coefficients based on the real-time ZTD of GNSS and ground meteorological stations on the average elevation surface, and obtain the real-time ZTD model of the study area.
Citation Information
Patent Citations
Real-time water vapor field generation method coupling GNSS and ground meteorological station monitoring data
CN115097490A
Troposphere delay modeling method fusing spherical harmonic function and fine grid
CN116029177A
