Satellite microwave humidity data all-weather assimilation method based on ARMS

Abstract

The invention relates to an ARMS-based satellite microwave humidity data all-weather assimilation method. The method comprises the following steps: acquiring YH4DVAR global background field data; calculating the effective particle radius of the hydrogel; establishing a symmetric error model aiming at the scattering characteristics of the ARMS for the forecast deviation of the cloud and rainfall intensity and the spatial position, expanding the observation error into a piecewise function of the cloud amount or the rainfall amount in the mode and observation by the symmetric error model, and distributing assimilation weights for microwave observation data under the all-weather condition; and assimilating and forecasting all-weather satellite microwave humidity data by using the model. The method has a good all-weather microwave radiation simulation capability, provides a new observation operator for YH4DVAR all-weather assimilation, and has a wide application prospect in the field of satellite microwave data assimilation.

Description

Technical Field

[0001] This invention belongs to the field of weather forecasting technology, and in particular relates to an all-weather assimilation method for satellite microwave humidity data based on ARMS. Background Technology

[0002] Typhoons are intense cyclonic vortex systems that form in the tropical ocean atmosphere. They are typically accompanied by severe weather phenomena such as strong winds and heavy rainfall, and are among the most threatening natural disasters to human activities. Accurate numerical weather prediction (NWP) can provide precise forecasts of the path and intensity development of extreme weather events such as typhoons. Satellite microwave sounding data is an important observational data source for NWP. Studies have shown that microwave sounders can penetrate non-precipitating clouds, providing temperature and humidity profiles for NWP models, greatly improving the accuracy of short- and medium-range weather forecasts and typhoon forecasts.

[0003] Fast Radiative Transfer (RTTOV) models serve as a bridge between model space and satellite observation space, and are a key technology for achieving all-weather microwave data assimilation. Under all-weather conditions, radiative transfer simulation requires not only conventional model variables (temperature, air pressure, humidity, wind) but also the additional consideration of the scattering characteristics of various hydrophobic substances (water clouds, ice clouds, rain, snow, graupel, hail), making radiative transfer calculations under scattering conditions extremely complex. Existing all-weather microwave data assimilation schemes typically use the RTTOV_SCATT scattering module in the Fast Radiative Transfer (RTTOV) model as the observation operator to simulate microwave radiation data under cloud and water conditions. Regarding hydrophobic particle shape, hydrophobic substances from water clouds, ice clouds, rain, and snow are usually assumed to be spherical particles, and their corresponding optical properties (attenuation coefficient, single-scattering albedo, and asymmetry) are pre-calculated and stored in the Mie-table for quick retrieval during calculations. For radiative transfer solutions, the RTTOV-SCATT multiple scattering solution uses the Eddington approximation, which is essentially a two-stream approximation spherical harmonic function solution.

[0004] ARMS (Advanced Radiative Transfer Modeling System) is a rapid radiative transfer model developed by the China Meteorological Administration. In the cloud and precipitation scattering module, ARMS constructs a non-spherical ice-phase particle scattering lookup table (LUT) for the microwave band. Liquid particles (cloud water and rain) are still handled based on Mie sphere theory, while ice-phase particles (cloud ice, snow, and graupel) are uniformly handled using a roughened six-bullet bouquet aggregate model, and the discrete dipole approximation (DDA) is used to calculate the optical properties of single particles. For radiative transfer solutions, ARMS introduces the Accelerated Discrete Ordinate Method (ADOM) to improve the efficiency of multi-layer multi-scattering calculations. ADOM is a multi-flow radiative transfer solution scheme that can distinguish atmospheric layers according to scattering / non-scattering characteristics, significantly reducing matrix dimensions and computational costs. It is important to note that when calling the ARMS scattering module, users are required to input the effective particle radius for various hydrophenate types. Users need to calculate the corresponding effective particle radius based on their own cloud microphysical parameterization scheme to make ARMS adaptable to various cloud microphysical parameterization schemes, increasing the difficulty of using ARMS. Summary of the Invention

[0005] This application utilizes the YH4DVAR assimilation system developed by the National University of Defense Technology, employing the domestically developed Rapid Radiative Transfer Mode (ARMS) as the observation operator to achieve all-weather assimilation of the MetOp-C MHS. First, considering the characteristics of the cloud microphysical parameterization scheme in YH4DVAR, the effective particle radius of condensates was calculated. Second, a symmetric error model for the scattering characteristics of ARMS was established. Finally, a forecast experiment was conducted using Typhoon "Mangkut" as a case study. Through comparative analysis with RTTOV, the assimilation effect and forecasting performance of ARMS were systematically evaluated. Experimental results show that: (1) Under all-weather conditions, the OB (observed brightness temperature radiation minus the simulated brightness temperature of the background field) of this application is basically equivalent to that of RTTOV. The mean OB values ​​of ARMS over the ocean in channel 4 (183±3 GHz) and channel 5 (190.31 GHz) are 0.05 K and -0.29 K, respectively, which are slightly better than RTTOV (-0.35 K, -0.47 K); (2) Compared with RTTOV11.2, the wind path prediction error is reduced by about 13% and the minimum pressure error is reduced by about 6% when using this application. This application has good all-weather microwave radiation simulation capabilities, providing a new observation operator for YH4DVAR all-weather assimilation, and has broad application prospects in the field of satellite microwave data assimilation.

[0006] To achieve the above objectives, this application discloses an all-weather assimilation method for satellite microwave humidity data based on ARMS, comprising the following steps: Collect YH4DVAR global background field data; Calculate the effective particle radius of the hydrogel; To address the forecast biases of cloud and precipitation intensity and spatial location, a symmetric error model for ARMS scattering characteristics is established. This symmetric error model extends the observation error into a piecewise function of cloud amount or precipitation in the model and the observation, and assigns assimilation weights to microwave observation data under all-weather conditions. The model is used to assimilate and forecast all-weather satellite microwave humidity data.

[0007] Furthermore, the effective particle radius Represented as: ; Where r is the particle radius. It is a particle spectrum distribution with radius as the variable. These are shape parameters. It is the mixing ratio of hydrogel x. It is the density of the hydrogel. N is the density of air, and N is the particle number concentration. It is a gamma function; Mixing ratio of hydrogel Defined as: ; in, It is the density of the hydrogel. It is the density of air. Indicates the distribution of cloud particle spectra. For the intercept parameter, It is the diameter of cloud particles. Indicates the slope parameter; Air density is expressed as: ; in , , These represent the air pressure, specific humidity, and temperature at the level, respectively. slope for: .

[0008] Furthermore, the effective particle radius of water clouds in terrestrial and marine hydrophobic condensates... and They are respectively: ; ; Among them, the number concentration of terrestrial water cloud particles Oceanic water cloud particle number concentration , It is the density of water clouds. It is the water-cloud mixing ratio.

[0009] Furthermore, the formula for the effective particle radius of ice clouds in hydrogels is: ; Among them, ice cloud density , It is the number concentration of ice cloud particles. It refers to the ice cloud content.

[0010] Furthermore, the effective particle radius of rain particles in hydrogels The calculation formula is: ; Where a and b are coefficients, for rain particles a = π / 6, b = 3.0, and These are the coefficients for rain particles. =0.22, =2.2.

[0011] Furthermore, the effective particle radius of snow particles in hydrogels The calculation formula is ; in, Freezing point temperature Atmospheric temperature, It is the mixing ratio of snow particles, where a and b are coefficients. For snow particles, a = 0.069 and b = 2.0.

[0012] Furthermore, the symmetric error model is shown in the following equation: ; In the formula, For symmetric cloud prediction operators, , , , These represent the observation errors under clear sky conditions, the observation errors under cloud and rain conditions, and the minimum and maximum values ​​of the symmetric cloud prediction operator, respectively. Attached Figure Description

[0013] Figure 1 This is a flowchart of the assimilation method provided in the embodiments of this application. Detailed Implementation

[0014] To further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following specifically describes, with reference to the accompanying drawings and preferred embodiments, the specific implementation manners, structures, features and effects of the present invention as follows.

[0015] This application first introduces the data types used in the field of data assimilation: 1. Microwave Humidity Sounder (MHS) is a five-channel microwave radiometer carried on the Metop-C satellite. MHS adopts a cross-track scanning method. Each scanning line contains 90 detection pixels. The ground resolution is 17 km. The width of each scanning line is about 2348 km. The scanning angle corresponding to the pixel at the edge of the orbit is 49.5°. Currently, all-weather assimilation of MHS has been achieved based on RTTOV_SCATT, effectively improving the humidity field.

[0016] 2. ARMS and RTTOV11.2: RTTOV11.2 is a fast radiative transfer model developed by the European EUMETSAT NWP-SAF and is an important observation operator for realizing all-weather assimilation of satellite microwave data. ARMS is a fast radiative transfer model developed by the China Meteorological Administration, mainly composed of forward, tangent linear and adjoint models. It can accurately simulate the brightness temperature of satellite infrared and microwave detection channels under clear sky conditions and has been successfully integrated into YH4DVAR.

[0017] In terms of cloud and precipitation scattering, ARMS constructs a look-up table (LUT) for non-spherical ice-phase particle scattering for the microwave band. Among them, liquid particles (cloud water and rain) are still processed based on Mie theory, while ice-phase particles (cloud ice, snow, graupel) uniformly adopt the roughened six-bullet bouquet aggregate model, and the discrete dipole approximation (DDA) is used to calculate the optical properties of single particles. Compared with the spherical assumption, this LUT significantly improves the scattering characterization in the middle and high-frequency microwave channels, especially in the case of deep convection, which is more consistent with the observations.

[0018] 3. YH4DVAR: YH4DVAR is a four-dimensional variational assimilation operational system developed by the National University of Defense Technology. This system uses the global spectral model as the dynamic balance constraint and uses a 12-hour assimilation time window to improve the accuracy of the atmospheric initial value. YHGSM is a global atmospheric spectral model配套 with YH4DVAR. The resolution of this model is about 16 km, and a total of 137 layers are set in the vertical direction, extending upward from the ground surface to 0.01 hPa. At the operational application level, this system has both clear sky assimilation and all-weather assimilation functions.

[0019] The ARMS scattering module requires input of condensate content and effective particle radius to retrieve the scattering lookup table and then calculate atmospheric transmittance under scattering conditions. In the YH4DVAR system, condensate types include four categories: water clouds, ice clouds, rain, and snow. The mass mixing ratio of condensate is a direct model forecast and can be directly used for radiative transfer calculations. However, the effective particle radius is not a forecast variable in YH4DVAR. The corresponding effective particle radius needs to be derived based on the characteristics of each condensate parameterization scheme to meet the ARMS scattering module's input requirement for the effective particle radius.

[0020] In one embodiment, the formulas for calculating various effective particle radii are as follows: Assuming cloud particles follow a gamma distribution, then: (1); in Cloud particle spectral distribution (unit: m) -3 m -1 ), For the intercept parameter, It is the diameter of cloud particles. These are shape parameters. This represents the slope parameter. By integrating, the particle number concentration can be obtained. (Unit: m) -3 ); (2); To construct the gamma function, let ,but The above formula can be transformed into: (3); According to the definition of the gamma function: The above formula can be simplified to: (3); Effective particle radius Defined as: (4); Where r is the particle radius. This is the particle spectrum distribution with radius as the variable. Considering the particle diameter D=2r, then the particle spectrum distribution with diameter as the variable is... It can be transformed into a spectral distribution with radius as the variable. Then the integral It can be transformed into: (5); To construct the gamma function, we can let ,but The above formula can be transformed into: (6); According to the definition of the gamma function: The above formula can be simplified to: (7); Similarly, we can conclude that: (8); Therefore, the effective particle radius It can be represented as: (9).

[0021] The mixing ratio of hydrogel (cloud water, cloud ice, rain, snow) (kg kg) -1 The definition of ) is: (10); in, It is the density of the hydrogel. It is the density of air, according to the ideal gas law: in q is the specific humidity (unit: kg / kg). Therefore, air density can be expressed as: (11); in , , These represent the air pressure, specific humidity, and temperature at the level, respectively.

[0022] Particle number concentration Substituting into equation (10), then This can be expressed as: (12); So the slope for: (13); Will Substituting back into formula (9) yields the effective particle radius. The calculation formula is as follows: (14).

[0023] (2) Calculation scheme for effective particle radius of water cloud In YH4DVAR, the shape parameters of water clouds ,density Particle number concentration It is a constant, but it varies on land and at sea. , Substituting into formula (14), the effective particle radius on land and sea can be obtained. and They are respectively; (15); (16).

[0024] (3) Calculation scheme for effective particle radius of ice cloud: Ice cloud density in YH4DVAR Particle number concentration and ice cloud content They satisfy an approximate relationship: ,Will Substituting into formula (14), we obtain the formula for the effective particle radius of the ice cloud: (17).

[0025] (4) Calculation scheme for effective particle radius of rain particles: In YH4DVAR, there are specific approximate relationships (18)-(19) in the rain particle microphysics parameterization scheme. 1) Approximate relationship between particle mass and diameter: particle mass and diameter The relationship can be expressed by the following empirical formula: (18); Where a and b are coefficients, for rain particles a=π / 6, b=3.0; 2) Spectral distribution of rain particles under MP distribution conditions: Particle number concentration (Unit: m) -3 )for ; in It is the slope of the particle-scale distribution. It is the intercept parameter; It can be approximated as The function, intercept parameter and slope parameters Functional relationship: (19); in It is the slope of the particle-scale distribution. It is the intercept parameter, where and These are the coefficients for rain particles. =0.22, =2.2, mixing ratio (Unit: kg / kg) can be expressed as: (20); To construct the gamma function, let The above formula can be transformed into: (twenty one); The constructed gamma function Substituting into the above equation, we get: (twenty two); Will Substituting into the above formula, then the slope parameter It can be represented as: (twenty three); Will Substituting the expression into formula (9), the effective particle radius of the rain particles is... The calculation formula is: (twenty four).

[0026] (5) Calculation scheme for effective particle radius of snow particles Snow particles also follow an MP distribution with a = 0.069 and b = 2.0. In the YH4DVAR parameterization scheme, the intercept parameters of the snow particles are assumed to be... The following linear relationship exists between it and atmospheric temperature: (25); in, 273.15K is the freezing point temperature. Let be the atmospheric temperature. Substituting the above formula into equation (22), we can obtain the mixing ratio of the snow particles. (Unit: kg) can be expressed as: (26); slope parameter It can be represented as: (27); in, 273.15K is the freezing point temperature. This refers to the atmospheric temperature. Substituting the expression into formula (9), the effective particle radius of the snow particle is... The calculation formula is: (28).

[0027] This application, based on global background field data from YH4DVAR at 00:00 (UTC) on September 3, 2024, statistically analyzed the effective particle radius distribution of four types of condensates: water clouds, ice clouds, rain, and snow. The effective particle radius of water clouds is concentrated in the range of 5-15 μm. This is because, in the atmosphere, the growth of the effective particle radius of water clouds mainly relies on the collision and coalescence effect of cloud droplets. Its radius growth rate is inversely proportional to the current radius. When cloud droplet particles are smaller than 5 μm, the cloud droplet radius will rapidly grow and exceed 5 μm. However, when the cloud droplet particle radius grows beyond 15 μm, its growth rate will significantly decrease. The 5-15 μm range is a relatively stable interval, therefore most cloud droplet particles fall within this range. The effective particle radius distribution of ice clouds is concentrated in the range of 50-80 μm. This is because ice crystals mainly grow through water vapor sublimation. In environments where the ice surface is supersaturated and the water surface is unsaturated, ice crystals can rapidly grow to approximately 60 μm by absorbing surrounding water vapor (or being supplied by the evaporation of supercooled water droplets). When ice crystals exceed 80-100 μm in size, their terminal settling velocity increases significantly, gradually moving them out of the supersaturation zone, thus confining the particle size to a relatively stable range of 50-80 μm. Rain particles exhibit more complex and diverse characteristics (light rain, moderate rain, heavy rain, torrential rain, extremely heavy rain), therefore their effective particle radius varies from approximately 50-800 μm, with the corresponding particle quantity decreasing progressively from high to low. The distribution characteristics of snow particles are similar to those of rain particles.

[0028] Within the all-weather assimilation framework, observation errors typically include instrument noise, radiative transfer simulation errors, and model forecast biases for cloud and precipitation intensity and spatial location. Specific observation error models need to be constructed to characterize these errors, thereby allocating reasonable assimilation weights to observation data under different environmental conditions.

[0029] This application uses a symmetric error model to describe this error, which extends the observation error to a function of symmetric cloud cover (the average of the cloud cover or precipitation in the model and the observations). For observation data from the MetOp-C MHS microwave hygrometer, the scattering index SI is used as the method for calculating cloud cover. The scattering index SI is calculated as the difference between the brightness temperature value at 90 GHz, which is less affected by cloud and rain scattering, and the brightness temperature value at 150 GHz, which is more affected by cloud and rain scattering. For the scattering index over land, it can be expressed as: ; SI can represent the scattering of water condensate particles in clouds, and the corresponding symmetric cloud prediction operator can be expressed as: ; In the formula , and These are, respectively, a symmetric cloud prediction operator, a scattering index calculated from observations, and a scattering index calculated from the background field. This operator can effectively reduce the representativeness errors caused by the difference between the background field calculated from the radiative transfer model and the observational data in terms of the time, intensity, and location of cloud and rain occurrence.

[0030] Because the scattering index SI is sensitive not only to scattering but also to water vapor absorption over the sea surface, the effect of water vapor absorption must be removed when calculating the scattering index. ): ; This application uses a symmetric error model to describe the observation error, which extends the observation error to a function of symmetric cloud cover (the average of cloud cover or precipitation in the model and the observations), as shown in the following equation; ; In the formula , , , These represent the observation errors under clear sky conditions, the observation errors under cloud and rain conditions, and the minimum and maximum values ​​of the symmetric cloud prediction operator. Between the thresholds of the two symmetric cloud prediction operators, a quadratic function is used to describe the relationship between the observation error and the symmetric cloud operator.

[0031] The following are the results of the analysis of the relationship between the standard deviation of observable clouds (OB) and the symmetric cloud scattering factor for MetOp-C MHS channels 3-5 in September 2024, statistically analyzed over land and ocean areas: Overall, the OB standard deviation gradually increases with the increase of the symmetric cloud scattering factor. Quadratic fitting and piecewise methods can fit the relationship well. The lower the extreme value of the channel weight function, the more obvious the influence of the underlying surface, and the larger the OB standard deviation. Therefore, under the same symmetric cloud predictor factor, the order of OB standard deviation from largest to smallest is: channel 5 (190 GHz), channel 4 (183 ± 1.0), and channel 3 (183 ± 3.0).

[0032] Typhoon Mangkhut, the 11th typhoon of 2024, entered the South China Sea on September 2 and rapidly intensified, reaching its strongest intensity in its lifetime on September 6. It then made landfall on the coast of Hainan Island, China. The typhoon had a complete circulation structure and extremely strong intensity, causing serious impact on southern China.

[0033] This application designed two sets of cyclic assimilation experiments, which assimilated the all-weather radiative brightness temperature of conventional observations and Metop-C MHS, respectively. The difference between the two sets of experiments lies in the fast radiative transfer mode. Test_arms uses the ARMS fast radiative transfer mode, while Test_rttov uses RTTOV11.2's RTTOV_SCATT.

[0034] (1) OB analysis: An image of the OB distribution in MHS channel 4 (183.31±3.0 GHz) within the region of Typhoon Mangkhut was acquired at 00:00 (UTC) on September 3, 2024. This image reveals the macroscopic structural features of the typhoon, including a clear eye (around 19°N, 117°E), eyewall, and peripheral spiral rainbands, serving as a reference for typhoon morphology. The results show that in clear-sky areas, the brightness temperature residuals simulated by both models are close to zero and their spatial distributions are almost identical, with a difference of less than 0.5 K, demonstrating good consistency. However, in the strong convective cloud and rain areas such as the typhoon eyewall and spiral rainbands, the OB difference between RTTOV and ARMS shows a positive deviation of approximately +1.5 K. This difference indicates that ARMS has a stronger scattering effect on condensate particles, resulting in more radiant energy being scattered.

[0035] This application statistically analyzes the OB characteristics of Metop-C MHS data entering the numerical weather prediction assimilation system globally for channels 3 (H3, 183.31±1.0 GHz), 4 (H4, 183.31±3.0 GHz), and 5 (H5, 190.31 GHz). Considering that MHS channels 1 (89 GHz) and 2 (157 GHz) are window channels primarily used for precipitation detection and cloud area identification and do not directly enter the assimilation system, the statistics only focus on water vapor channels 3-5 that participate in the assimilation. The effective sample numbers for RTTOV over land and ocean are 47,297 and 195,417, respectively, and the ARMS are 48,152 and 191,650, respectively; the slight difference in sample numbers between the two groups mainly stems from slight differences in the quality control data removal.

[0036] Overall, the absolute values ​​of the mean OB values ​​for both radiative transfer modes (RTTOV and ARMS) are less than 0.5 K, and the standard deviations are within the range of 2.0-2.6 K with little difference between them, both meeting the data quality standards for microwave humidity data in numerical weather prediction assimilation systems. Specifically, over the ocean, ARMS has mean OB values ​​of 0.05 K and -0.29 K in channel 4 (183±3 GHz, sensitive to mid-level humidity) and channel 5 (190.31 GHz, sensitive to low-level humidity), respectively, which are closer to zero than RTTOV's -0.35 K and -0.47 K, showing an advantage in simulating mid-to-low-level water vapor radiative transfer. Over land, ARMS performs slightly better than RTTOV in bias control in channel 4, but slightly worse than RTTOV in channel 5.

[0037] Standard deviation statistics show that both models exhibit a decreasing trend with decreasing detection altitude (Channel 3 > Channel 4 > Channel 5), which corresponds precisely to the peak height of the weighting function for the corresponding channel. This is mainly due to the differences in sensitivity to atmospheric parameters and background field error characteristics among different channels: Channel 3 (183±1 GHz) is located in the wing region of the strong water vapor absorption line, and the peak of the weighting function is located in the upper troposphere. Although the water vapor content in this region is low, the vertical gradient is large, and the prediction error of upper-layer humidity by numerical models is usually large. The uncertainty of the cloud microphysics scheme leads to an increase in OB dispersion. Channel 5 (190.31 GHz) is relatively far from the center of the water vapor absorption line, and the peak of the weighting function is located in the lower troposphere. This region has abundant water vapor, and the background field humidity prediction is relatively accurate, thus exhibiting a smaller standard deviation. The statistical characteristics of Channel 4 (183±3 GHz) are between the two, reflecting the transitional characteristics of the mid-level atmospheric humidity gradient and scattering effect.

[0038] (2) Analyze the increment: The incremental distribution of the 500 hPa geopotential height field was analyzed after all-weather assimilation of the MetOp-C MHS using both RTTOV and ARMS radiative transfer modes. The spatial distribution of the geopotential height increments generated by the two modes is highly consistent, with the lowest geopotential height at 573 dagpm in the closed low-pressure center of Typhoon Mangkhut's region in both models. Comparing the difference fields between the two, the increment differences are generally close to zero, and the spatial distribution exhibits weak random noise characteristics without systematic bias.

[0039] After collecting all-weather assimilation of Metop-C MHS data using RTTOV and ARMS radiative transfer modes, the analytical increment distributions of the 850 hPa temperature field (upward) and relative humidity field (downward) were analyzed. Temperature field analysis showed that the increments generated by the two modes exhibited high spatial consistency, with the centers of positive and negative disturbances largely matching in location and intensity. This is because the MHS, as a microwave humidity sounder, is sensitive to changes in atmospheric humidity, but its direct adjustment effect on the temperature field is relatively limited; therefore, the differences in temperature analysis increments between different radiative transfer modes are weakened. In contrast, the MHS, as a microwave humidity sounder, is sensitive to atmospheric water vapor content in its 183 GHz channel, thus its impact on the relative humidity field increments after all-weather assimilation is more significant. Notably, in the core area and spiral rainband region of Typhoon Mangkhut, there are significant differences in the relative humidity increments generated by ARMS and RTTOV, with deviations reaching ±5%. This difference may stem from two aspects: First, the radiative transfer solution schemes are different. RTTOV-SCATT uses the Eddington approximation (two-stream approximation), while ARMS is based on the Accelerated Discrete Coordinates Method (ADOM, multi-stream approximation). The difference in the number of streams may lead to different scattering calculation accuracy. Second, the particle scattering model assumptions are different. RTTOV simplifies all hydrophobic particles as spherical particles, while ARMS assumes that liquid particles (cloud water and rain) are spherical particles. For ice cloud and snow cloud phase particles, it uses a roughened six-bullet bouquet aggregate model combined with the Discrete Dipole Approximation (DDA) method to simulate the hydrophobic scattering effect. This difference in microphysics models may affect the simulation of microwave scattering, and thus be transferred to the humidity field through assimilation.

[0040] 3) Typhoon path and intensity analysis: Based on the all-weather assimilation of Metop-C MHS data using the ARMS and RTTOV radiative transfer models, the track forecast performance of Typhoon Mangkhut (reported from 00:00 on September 3, 2024) was verified. Within the first 0-72 hours, the simulation results of the two models were generally similar, with the ARMS forecast track being closer to reality than the RTTOV forecast. After 72 hours, the typhoon made landfall, and due to the complex underlying surface (topographic friction, land-sea thermal differences) and adjustments in the environmental flow field, the uncertainty of the typhoon track forecast increased significantly. Track error statistics show that the error growth of both models was relatively gradual before 72 hours; after 72 hours, as the typhoon made landfall in Hainan, the track errors of both models increased significantly due to the aforementioned complex physical processes. Throughout the 0-120 hour forecast period, the average track error of the ARMS all-weather assimilation scheme was 137 km, significantly reduced by approximately 21 km compared to the RTTOV's 158 km, representing a reduction of approximately 13% in the typhoon track error.

[0041] The time-lead time evolution of Typhoon Mangkhut intensity forecast error based on all-weather assimilation of Metop-C MHS data using the ARMS and RTTOV radiative transfer models. The forecast errors for the minimum pressure of the typhoon in both models initially increased and then decreased with forecast lead time, reaching a peak at 48 hours (approximately 65 hPa). Thereafter, the errors gradually decreased with the typhoon's landfall and structural changes. Statistical analysis shows that the ARMS average pressure forecast error was 28.56 hPa, approximately 1.83 hPa less than that of RTTOV, representing a reduction of about 6%. The wind speed error evolution exhibited similar characteristics, with a maximum negative deviation (approximately -37 m / s) appearing around 48 hours. The ARMS average wind speed error was essentially equivalent to that of RTTOV.

[0042] The beneficial effects of this application are as follows: Compared with the traditional method of using RTTOV software to achieve all-weather assimilation of satellite microwave humidity data, this application is the first to achieve all-weather assimilation of microwave humidity data based on the domestic fast radiative transfer mode ARMS, and the effect is better than the RTTOV method.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for all-weather assimilation of satellite microwave humidity data based on ARMS, characterized in that, Includes the following steps: Collect YH4DVAR global background field data; Calculate the effective particle radius of the hydrogel; To address the forecast biases of cloud and precipitation intensity and spatial location, a symmetric error model for ARMS scattering characteristics is established. This symmetric error model extends the observation error into a piecewise function of cloud amount or precipitation in the model and the observation, and assigns assimilation weights to microwave observation data under all-weather conditions. The model is used to assimilate and forecast all-weather satellite microwave humidity data.

2. The all-weather assimilation method for satellite microwave humidity data based on ARMS according to claim 1, characterized in that, Effective particle radius Represented as: ； Where r is the particle radius. It is a particle spectrum distribution with radius as the variable. These are shape parameters. It is the mixing ratio of hydrogel x. It is the density of the hydrogel. N is the density of air, and N is the particle number concentration. It is a gamma function; Mixing ratio of hydrogel Defined as: ； in, It is the density of the hydrogel. It is the density of air. Indicates the distribution of cloud particle spectra. For the intercept parameter, It is the diameter of cloud particles. Indicates the slope parameter; Air density is expressed as: ； in , , These represent the air pressure, specific humidity, and temperature at the level, respectively. slope for: 。 3. The all-weather assimilation method for satellite microwave humidity data based on ARMS according to claim 2, characterized in that, Effective particle radius of water clouds in condensates on land and sea and They are respectively: ； ； Among them, the number concentration of terrestrial water cloud particles Oceanic water cloud particle number concentration , It is the density of water clouds. It is the water-cloud mixing ratio.

4. The all-weather assimilation method for satellite microwave humidity data based on ARMS according to claim 2, characterized in that, The formula for the effective particle radius of ice clouds in hydrogels is: ； Among them, ice cloud density , It is the number concentration of ice cloud particles. It refers to the ice cloud content.

5. The all-weather assimilation method for satellite microwave humidity data based on ARMS according to claim 2, characterized in that, Effective particle radius of rain particles in hydrogel The calculation formula is: ； Where a and b are coefficients, for rain particles a = π / 6, b = 3.0, and These are the coefficients for rain particles. =0.22, =2.

2.

6. The all-weather assimilation method for satellite microwave humidity data based on ARMS according to claim 2, characterized in that, Effective particle radius of snow particles in hydrogel The calculation formula is ； in, Freezing point temperature Atmospheric temperature, It is the mixing ratio of snow particles, where a and b are coefficients. For snow particles, a = 0.069 and b = 2.

0.

7. The all-weather assimilation method for satellite microwave humidity data based on ARMS according to claim 1, characterized in that, The symmetric error model is shown in the following equation: ； In the formula, For symmetric cloud prediction operators, , , , These represent the observation errors under clear sky conditions, the observation errors under cloud and rain conditions, and the minimum and maximum values ​​of the symmetric cloud prediction operator, respectively.

Images