Numerical model dynamical framework construction method and apparatus for weather prediction

Abstract

The application relates to a numerical model dynamic framework construction method for weather prediction, which comprises the following steps: establishing a basic equation group of atmospheric movement of a numerical model dynamic framework and an energy conversion relationship; and based on the basic equation group of atmospheric movement of the numerical model dynamic framework and the energy conversion relationship, a dynamic framework space discrete model is established. The application significantly improves the energy balance property of the model, improves the ability of the model to depict fine structures, and improves the reaction ability of the model to the influence of terrain. The application also relates to an equipment and a storage medium.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of weather forecasting, and particularly relates to a numerical model dynamic framework construction method and device for weather forecasting. BACKGROUND

[0002] With the continuous improvement of computer technology and the progress of diverse and refined observation technology, the spatiotemporal resolution of numerical weather prediction models has been significantly improved, and cloud-resolving scale numerical models have been widely used in the simulation of severe convective weather forecasting. However, the existing numerical weather prediction models still have low prediction ability for severe convective weather, and the problems in the discrete scheme of the model dynamic framework may be one of the fundamental reasons for the low prediction ability of the current numerical models for severe convective weather. The model dynamic framework is closely related to the non-adiabatic physical processes of the model, and its defects will inevitably affect the effects of other physical processes.

[0003] The occurrence and development of severe convective weather are not only affected by large-scale weather systems, but also involve the combination or merging process of small-scale systems, and the multiscale interaction is the fundamental reason for the generation of severe convective weather. The inability of numerical models to reproduce the multiscale interaction process, especially the effective simulation of the combination or merging process of small-scale systems, is the main reason for the low level of current severe convective weather prediction. Therefore, improving the description of multiscale interaction is one of the difficult problems in constructing numerical weather prediction models. Multiscale interaction is directly manifested as the generation, disappearance and transformation of wind pressure fields of various scales, and essentially represents the scale change of kinetic energy and potential energy, i.e., the dispersion characteristics of energy. SUMMARY

[0004] In view of the above problems existing in the current numerical weather prediction model dynamic framework, the present application provides a numerical model dynamic framework construction method and device for weather forecasting.

[0005] In a first aspect, the present application provides a numerical model dynamic framework construction method for weather forecasting, which comprises:

[0006] establishing a basic equation set of atmospheric motion and an energy conversion relationship of the numerical model dynamic framework;

[0007] based on the basic equation set of atmospheric motion and the energy conversion relationship of the numerical model dynamic framework, establishing a dynamic framework spatial discrete model.

[0008] In a second aspect, the present application further provides a numerical model dynamic framework construction device for weather forecasting, which comprises:

[0009] a first processing module, configured to establish a basic equation set of atmospheric motion and an energy conversion relationship of the numerical model dynamic framework;

[0010] The second processing module is configured to establish a dynamic framework space discrete model based on the basic equation set of atmospheric movement of the numerical model dynamic framework and the energy conversion relationship.

[0011] In a third aspect, the present application provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method for constructing a numerical model dynamic framework for weather forecast according to any one of the first aspect.

[0012] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method for constructing a numerical model dynamic framework for weather forecast according to any one of the first aspect.

[0013] The method for constructing a numerical model dynamic framework for weather forecast provided by the present application comprises the following steps: establishing a basic equation set of atmospheric movement of a numerical model dynamic framework and an energy conversion relationship; and establishing a dynamic framework space discrete model based on the basic equation set of atmospheric movement of the numerical model dynamic framework and the energy conversion relationship. The present application significantly improves the energy balance property of the model, improves the ability of the model to depict fine structures, and improves the ability of the model to respond to the influence of terrain. BRIEF DESCRIPTION OF DRAWINGS

[0014] The accompanying drawings, which form a part of the present application, are intended to provide further understanding of the present application, and are used to interpret the illustrative embodiments of the present application and their descriptions, and do not constitute improper limitations to the present application. In the drawings:

[0015] Figure 1 FIG. 1 is a flowchart of the method for constructing a numerical model dynamic framework for weather forecast provided by an embodiment of the present application;

[0016] Figure 2 FIG. 2 is a topological diagram of a dynamic framework space discrete scheme provided by another embodiment of the present application;

[0017] Figure 3 FIG. 3 is a result diagram of a 100m resolution thermal bubble experiment provided by another embodiment of the present application;

[0018] Figure 4 FIG. 4 is a temperature profile diagram of a thermal bubble experiment provided by another embodiment of the present application;

[0019] Figure 5 FIG. 5 is a maximum temperature and maximum minimum wind speed time variation curve diagram of a thermal bubble provided by another embodiment of the present application;

[0020] Figure 6is a schematic diagram of an energy curve of a 100m FM and WRF warm bubble experiment provided by an embodiment of the present application;

[0021] Figure 7 is a schematic diagram of a potential temperature plane of a cold bubble of a 50m resolution Fm and WRF experiment provided by another embodiment of the present application;

[0022] Figure 8 is a schematic diagram of a potential temperature plane of a cold bubble of a 100m resolution Fm and WRF experiment provided by another embodiment of the present application;

[0023] Figure 9 is a schematic diagram of a potential temperature plane of a cold bubble of a 50m resolution Fm and WRF experiment provided by another embodiment of the present application;

[0024] Figure 10 is a schematic diagram of a potential temperature profile of a cold bubble of a 100m resolution FM and WRF experiment provided by another embodiment of the present application;

[0025] Figure 11 is a schematic diagram of a minimum potential temperature value and maximum and minimum wind speed values of a 100m resolution FM and WRF cold bubble experiment provided by an embodiment of the present application;

[0026] Figure 12 is a schematic diagram of an energy curve of a 100m FM and WRF cold bubble experiment provided by another embodiment of the present application;

[0027] Figure 13 is a schematic diagram of a WRF and FM mountain wave experiment provided by another embodiment of the present application;

[0028] Figure 14 is a schematic diagram of an energy diagnosis of a mountain wave experiment provided by another embodiment of the present application;

[0029] Figure 15 is a schematic diagram of a mountain wave vertical momentum flux provided by another embodiment of the present application;

[0030] Figure 16 is a schematic diagram of patent innovation provided by an embodiment of the present application;

[0031] Figure 17 is a schematic diagram of patent application scenarios provided by another embodiment of the present application;

[0032] Figure 18 is a schematic diagram of a flow of a numerical model dynamic framework construction method for weather forecasting provided by another embodiment of the present application;

[0033] Figure 19 is a schematic diagram of a numerical model dynamic framework construction module for weather forecasting provided by another embodiment of the present application. DETAILED DESCRIPTION

[0034] The present application will be described in detail below with reference to the attached drawings and embodiments. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0035] The following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise defined, all technical terms used in the present application have the same meanings as generally understood by those skilled in the art. The terms used in the present application are only for the purpose of describing the specific embodiments and are not intended to limit the exemplary embodiments according to the present application.

[0036] The factors affecting the prediction effect of numerical weather prediction model are generally summarized into three aspects: initial data processing, physical process scheme and dynamical core scheme. In comparison, the research on improving the simulation and prediction level of numerical model by improving the dynamical core scheme is slightly weak.

[0037] With the continuous improvement of computer technology and the progress of diverse and refined observation technology, the temporal and spatial resolution of numerical weather prediction model has been significantly improved, and cloud-resolving scale numerical model has been widely used in the simulation of severe convective weather. However, the existing numerical weather prediction model still has low prediction ability for severe convective weather, which is caused by many reasons. The problem of the discretization scheme of the dynamical core of the model may be one of the fundamental reasons for the low prediction ability of the numerical model for severe convective weather. The dynamical core of the model is closely related to the non-adiabatic physical processes of the model, and its defects will inevitably affect the effect of other physical processes. Constructing a discretization scheme of the dynamical core of the model suitable for describing severe convective weather processes will improve the prediction ability of the numerical model for severe convective weather.

[0038] The occurrence and development of severe convective weather are not only affected by large-scale weather systems, but also involve the combination or merging process of small-scale systems. Multi-scale interaction is the fundamental reason for the occurrence of severe convective weather. The inability of numerical models to reproduce the multi-scale interaction process, especially the effective simulation of the combination or merging process of small-scale systems, is the main reason for the low prediction level of severe convective weather. Therefore, improving the description of multi-scale interaction is one of the difficult problems in constructing numerical weather prediction models. Multi-scale interaction is intuitively represented by the birth, death and transformation of wind pressure fields of various scales, and essentially represents the scale change of kinetic energy and potential energy, i.e. the energy dispersion characteristics. Only by constructing a dynamical core discretization scheme that maintains the energy transmission and conversion characteristics, can the description ability of the numerical model for energy dispersion process be improved, and the numerical prediction level of severe convective weather be improved.

[0039] Generally, discrete equations cannot completely maintain the physical properties of the original continuous form equations, which can lead to computational disorder and nonlinear calculation instability. Thus, constructing a discrete scheme that can maintain the main physical properties of the continuous model directly affects the basic performance of the model dynamic framework and the model prediction ability. A good dynamic framework discrete scheme of a model not only maintains the overall physical properties of the original continuous form equations in the entire region, but also pays attention to maintaining the physical properties of each local space (such as maintaining the conversion relationship between total energy components in each microelement and the energy transmission relationship between adjacent microelements).

[0040] The representative numerical weather prediction models widely used at home and abroad include MM5, WRF, FV3, GRAPES, etc. Their dynamic framework spatial discrete schemes have different characteristics, but none of them considers the discrete construction from the perspective of maintaining the energy transmission and conversion characteristics in the local space. The MM5 dynamic framework discrete scheme does not consider maintaining the energy and mass properties in the non-hydrostatic sense, resulting in the discrete destruction of the related false sources and sinks of the energy overall properties and local allocation conversion mechanism. In order to maintain the stability of the calculation, numerical smoothing and dissipation techniques are excessively used in the model. These techniques have less damage to the long-wave part of the system and more damage to the medium and short-wave part, directly weakening the model's ability to simulate mesoscale weather processes. As an updated version of MM5, the WRF model starts to consider the energy and mass properties. Its dynamic framework better satisfies the mass conservation property; its momentum equation adopts a flux form, which better maintains the energy characteristics than the MM5 framework; due to the use of high-order upwind format for the advection term, the artificial viscosity dissipation term is reduced, the dissipation intensity of short waves and various scale components is reduced, and the model has better mesoscale weather simulation capability than MM5, and is currently widely used in numerical weather prediction simulation. However, the discrete form of the WRF dynamic framework still does not correctly reflect the local energy conversion relationship, and there are related false computational sources and sinks. FV3 is the dynamic framework scheme of the new generation of weather and climate integrated global prediction model (GFS) in the United States. Its characteristics are the use of scalable spherical cubed sphere grid, which improves the isotropy and local adaptability of the spherical grid and has high computational efficiency. Although it constructs the dynamic framework spatial discrete scheme from the perspective of finite volume, it does not consider the energy transmission relationship between grids, and there are related false computational sources and sinks. GRAPES is a numerical weather prediction model system independently developed by China, which is widely used in business prediction. Its dynamic framework scheme adopts a semi-Lagrangian scheme, which has high computational efficiency, but cannot maintain the energy spatial transmission and energy conversion relationship, and there are related false computational sources and sinks.

[0041] In view of the above-mentioned defects existing in the dynamic framework of the existing numerical weather prediction model, the application constructs a non-hydrostatic model dynamic framework spatial discretization scheme which keeps the energy transmission and conversion characteristics of local microelement, preferably depicts the mesoscale and small scale structure of the weather system, suppresses the weakening of the mesoscale and small scale weather process caused by excessive use of numerical smoothing and dissipation technology, and improves the prediction ability of the numerical model for strong convective weather process.

[0042] According to the energy conversion and transformation properties of the atmospheric motion equation set, the application constructs a dynamic framework spatial discretization scheme, and the given scheme can preferably keep the energy transmission and conversion characteristics of local microelement, thereby improving the ability of the model to depict the mesoscale and small scale structure of the weather system.

[0043] The application specifically realizes the steps as follows:

[0044] (1) Establish the atmospheric motion basic equation and its energy properties used in the dynamic framework scheme

[0045] As shown in Figures 1-2 With the increase of the resolution of the model grid, the energy cascade problem is highlighted in the simulation, especially the upward energy cascade becomes a prominent problem affecting the simulation effect. Satoh (2002, 2003, NICAM) constructs a dynamic framework scheme from the perspective of mass conservation and energy conservation, which can satisfy the conservation of mass, momentum and internal energy. However, due to the use of flux form momentum equation, it is difficult to effectively describe the kinetic energy transmission and conversion process, and it cannot satisfy the energy conservation property. A new energy fidelity model dynamic framework proposed in this paper adopts the WRF software architecture (the WRF Advanced Software Framework (ASF)), which aims to construct a dynamic framework from the perspective of satisfying the energy transmission and conversion process, effectively improve the calculation process of the energy cascade process of the model, and then improve the simulation effect of the high-resolution model. The new generation model dynamic framework proposed in this paper has two main purposes:

[0046] First, keep the transmission of each energy component and the conversion between them on the spatial grid microelement, especially reduce the error of kinetic energy transmission between grids; second, keep the energy transmission and conversion at grid scale, and keep the accuracy of the exchange between energy at various scales, which is beneficial to correctly describe the energy cascade process.

[0047] The basic equation set of atmospheric motion adopts the N-S equation set in the form of Lamb

[0048]

[0049] Where V=(u, v, w) is the atmospheric wind field, K is the atmospheric kinetic energy (K=|V| 2where p is the atmospheric density, p is the atmospheric pressure, T is the atmospheric temperature, F represents various dissipative forces, Q represents heat sources, q is the water vapor mixing ratio, S represents water vapor sources and sinks, Ω is the earth's rotation angular velocity, g is the gravitational acceleration, c v is the atmospheric specific heat at constant volume, and R is the atmospheric gas constant.

[0050] The above atmospheric motion equation set has the following energy conversion relationship

[0051]

[0052] where E = K + c v T + φ + Lq, c v T is the internal energy, φ is the gravitational potential energy Lq is the latent heat of water vapor.

[0053] When the system has no dissipation (F = 0) and only considers the heat change caused by phase change (Q = -LS), the energy conservation relationship at any point in space can be obtained,

[0054] 1) Atmospheric motion equation set under height terrain-following vertical coordinate

[0055] The horizontal coordinate of the atmospheric motion equation set is the local coordinate (x, y), and the vertical coordinate is the height terrain-following coordinate where z S = z S (x, y) is the ground height; z T is the top height of the atmospheric layer, which is taken as a constant.

[0056] In order to maintain the energy and mass properties, the basic prediction quantities are u, v, w, p, pT; map magnification factors m (x-direction grid spacing to real grid spacing in x-direction ratio), n (y-direction grid spacing to real grid spacing in y-direction ratio), here m, n are taken to be equal. The motion equation adopts a non-flux non-advection form, and the specific form is as follows

[0057] where

[0058]

[0059]

[0060]

[0061] where r = a + z (a is the earth's radius), is the latitude.

[0062] The above equation set under terrain coordinate satisfies the following energy conversion relationship

[0063]

[0064] (ii) Establishing the dynamic framework spatial discretization scheme

[0065] The discrete grid distribution of the forecast variables is Arakawa-C grid distribution horizontally and Lorenz distribution vertically. The uniform grid spacing Δx and Δy are adopted vertically, and the non-uniform grid spacing Δσ is adopted vertically.

[0066] The following discrete operators are defined

[0067]

[0068] The kinetic energy is defined at the Cross point (scalar point),

[0069] The principle of constructing the spatial discretization scheme is to satisfy the discrete form of the energy conversion relationship, and the premise for the discrete scheme to satisfy the energy flux expression form is that the energy conversion relationship is established. The most critical one is to satisfy the kinetic energy flux form, that is, the momentum equation (wind field) and the related discrete scheme of the continuity equation (density) satisfy the following kinetic energy flux relationship

[0070]

[0071] Based on the above constraint conditions, the following dynamic framework spatial discretization scheme can be obtained

[0072]

[0073]

[0074] 2) The dynamic framework spatial discretization scheme in the form of disturbance, which decomposes the state variables (ρ, T, p) into the basic

[0075] quantity (ρ b , T b , p b ) and the disturbance quantity (ρ’, T’, p’), that is

[0076] ρ=ρ b +ρ′, T=T b +T′, p=p b +p′

[0077] Assuming that the basic quantity only changes with height and satisfies the hydrostatic balance relationship The basic temperature is given by the average temperature vertical profile. The corresponding atmospheric motion equation in the form of disturbance is as follows

[0078]

[0079] The corresponding discrete form is

[0080] 3) Time Discretization:

[0081] The time discretization uses the classical semi-implicit (HE-VI) time-splitting algorithm, with the fast wave dependent terms integrated using a semi-implicit small time step and the rest of the terms integrated using a large time step. The following is the discretization scheme for the small time step (Δτ is the small time step and Δt is the large time step)

[0082]

[0083] Appendix: Surface Vertical Velocity

[0084]

[0085] Using the spatial discretization scheme of the dynamical framework and combining with the widely used semi-implicit time-splitting discretization scheme, a complete numerical weather prediction model dry framework is constructed, and a variety of numerical simulation ideal experiments are carried out. By comparing with the corresponding results of WRF model framework, obvious effects are obtained. Mainly reflected in the following three aspects:

[0086] 1) Significantly improve the energy balance properties of the model

[0087] 2) Improve the ability of the model to depict fine structures

[0088] 3) Improve the model's ability to respond to the impact of terrain

[0089] Example analysis:

[0090] Density current experiment

[0091] The most effective way to evaluate a numerical modeling system is to compare it with known results from cases analyzed dynamically, or with benchmark solutions that converge under certain conditions. Commonly used analysis and or benchmark cases include the mountain wave solution (e.g., Clark 1977; Dudhia 1993), the inertial-gravity wave (Skamarock and Klemp 1994), the nonlinear evolution of a cold pool and density current (Straka et al. 1993), and the rising warm current (Tripoli 1992). Simulations such as these are important for many reasons, such as establishing the fidelity of a new numerical modeling system, or testing the accuracy, efficiency, and efficacy of a new numerical technique

[0092] Using the newly established conservative model and the WRF model to conduct density current experiments, two-dimensional dry and cold bubble ideal numerical experiments are used to compare the differences between the experimental results. The two models are set with the same initial conditions, as follows:

[0093]

[0094] Table 1 Model parameter configuration

[0095] The verification of the mode uses a two-dimensional thermal bubble experiment in the density flow experiment. In a closed environment with a spatial resolution of 100x100m, a bubble with a diameter of about 50m is set at a height of 10m from the ground. The initial state of the air bubble is static, and the initial wind field is zero. The thermal bubble is in isentropic and static equilibrium. The potential temperature is initially 300K, and the initial surface air pressure is 1000mb. The bubble develops in an environment that is 0.5°C higher than the bubble, and is affected by buoyancy, causing the thermal bubble to rise and develop.

[0096] Figure 3 The selected figures are typical moments in the development process. The initial thermal bubble has a completely consistent initial field due to the consistency of the environmental conditions. In the process of thermal bubble development, it is affected by buoyancy and continuously rises and develops. In the process of rising, the thermal bubble stretches, and at the 10th moment, differences have appeared. The maximum center of the thermal bubble in the FM experiment is about 0.5K higher than that in the WRF experiment, and the shape is relatively consistent. At the 20th moment, the thermal bubble develops into a semi-ring structure, and the shape of the WRF and FM experiments is consistent. The maximum center of the thermal bubble in the FM experiment is about 1K higher than that in the WRF experiment. The overall potential temperature has a significant difference. At the last moment, the potential temperature of the thermal bubble in the WRF experiment is about 300.1-300.3K, while that in the FM experiment is 300.1-301.1K. It is obvious that the potential temperature of the thermal bubble in the FM experiment is higher. In the entire development process of the thermal bubble, the shape development of the Fm experiment and the WRF experiment is almost consistent, but the Fm mode shows a lower potential temperature decay rate, indicating that the energy fidelity is better than the WRF mode.

[0097] Figure 4 The figure shows the profile of the overall area-averaged potential temperature, which is used to observe the evolution of the thermal bubble experiment with height and time. The initial potential temperature profile is consistent. With the development of time, some differences appear between the FM mode and the WRF mode. At the 10th moment, the potential temperature of Fm is about 0.1 degrees higher than that of WRF mode. At the 20th moment, the potential temperature shows a large difference, mainly in the height of 40-60 layers. The maximum difference is that FM is about 0.5 degrees higher than WRF. At the last moment, the potential temperature of WRF is basically consistent with that of FM in the height of 60-90 layers. The potential temperature of WRF is higher than that of FM in the height of 90-110 layers. The maximum difference is obvious at the 120th layer, about 1 degree. In the entire development process of the thermal bubble, the maximum center potential temperature of the FM experiment is always close to the initial potential temperature value, while the potential temperature value of the WRF thermal bubble continuously decreases in the development process.

[0098] It is further proved that the FM mode has a smaller decrease in potential temperature than the WRF mode in the development of the thermal bubble.

[0099] Figure 5 As can be seen from the maximum potential temperature curve, the potential temperature of the WRF decreases continuously over time, while the maximum value of the FM decreases steadily and rebounds at the end. There is also a significant difference in the change of the maximum and minimum wind speed, and the maximum and minimum wind speed simulated by the FM is greater than that of the WRF mode.

[0100] As Figure 6 shown, further analysis of the energy diagnosis of the FM mode and the WRF mode is carried out. The figure is the change curve of the internal energy, kinetic energy, potential energy and total energy of the FM and WRF mode thermal bubble experiment in each development process. From the kinetic energy, the difference between the two is small, and the kinetic energy is continuously increasing due to the influence of buoyancy; from the potential energy curve, the potential energy of the WRF increases at 0-5, remains stable at 10-25 minutes, and slightly decays at the last stage, while the potential energy of the FM experiment continuously decreases in fluctuation; the change of internal energy is more obvious, the internal energy of the WRF experiment continuously and rapidly decays in the process of thermal bubble development, while the internal energy of the FM experiment has a smaller effective amplitude and tends to be stable in fluctuation. From the total energy curve, the energy line of the FM is almost straight, and the total energy of the WRF is attenuated, further proving the energy conservation of the FM mode.

[0101] From the energy conversion, the potential energy of the FM continuously decreases and converts into kinetic energy, and the internal energy also decreases and converts into kinetic energy, and at about 24 minutes, the decrease of potential energy exceeds that of internal energy. In the WRF experiment, the internal energy mainly converts into kinetic energy and part of the potential energy. This is also the obvious difference between the two modes.

[0102] The initial condition of the cold bubble is similar to that of the thermal bubble experiment, i.e. the four-wall rigid boundary condition, the initial u and v velocities are set to zero. Under hydrostatic equilibrium, the neutral atmosphere is initialized at 300K, and the grid resolution is set to three groups of control experiments 50m, 100 and 200m in the x and y directions respectively to study the difference of different resolutions on the simulation of cold bubble experiment, and the simulation running time is 900s, i.e. 15 minutes.

[0103] From high altitude to the ground, the temperature decreases linearly from 300K to 290K, which is like the opposite case of the previous thermal bubble, and the negative buoyancy makes the cold air bubble move downward, and when the cold bubble reaches the ground, a forward-propagating front is formed. This forward movement of the cold front produces shear at the top boundary of the front, thus generating Kelvin-Helmholtz waves. In the simulation, the propagation of the front and the formation of these waves can be seen at time=450s and time=900s Figure 3 ).

[0104] Figure 7As shown, the 50m resolution cold bubble experiment can be seen that the consistent initial field, at 5 minutes, the lowest center value of FM is 286 WRF, the lowest center is 288, the difference is about 2K, to 10 minutes, the shape of the cold bubble appears a big difference, the structure of the cold bubble Fm compared to WRF has a larger gradient, the shape also appears some differences, Kelvin-Helmholtz wave appears, at the last moment, Fm experiment compared to WRF appears clear wave train structure near the ground, Fm cold bubble has a lower cold center in the front of the wave train, while WRF experiment has a whole value greater than 290K, similar to the conclusion of the hot bubble experiment, FM experiment has better energy fidelity properties.

[0105] Figure 8 In the 100m resolution cold bubble experiment, the lowest center value of FM experiment is 286K at t=5min, and the lowest center value of WRF experiment is 288K, with a difference of about 2K. After t=10min, the shapes of the cold bubbles in the two experiments are different. The structure gradient of the cold bubble in FM is larger than that in WRF experiment, and the shape also has certain differences. WRF experiment shows that there is a clear wave train structure near the ground. FM experiment has a lower cold center in the front of the wave train, while the whole temperature in WRF experiment is >290K, which is similar to the result of the hot bubble experiment. FM experiment has better energy fidelity than WRF model.

[0106] Figure 9 As shown, the 200m resolution cold bubble experiment conclusion is similar to that of 50 and 100m resolutions. The cold center of the cold bubble becomes colder during the descent process, and the structure of the Kelvin-Helmholtz wave is different from WRF. The difference lies in that the shape of the Kelvin-Helmholtz wave of the 200m resolution cold bubble is different, the wave train structure is not as clear as that of low resolution, and the Kelvin-Helmholtz wave is more fragmented at the last moment and the lowest center temperature in the front is lower.

[0107] Figure 10 As shown, from the perspective of the potential temperature profile of the cold bubble, the difference between FM and WRF is small overall. At the 6th moment, there is an obvious difference in the potential temperature profile in the lower layer, but the profile in the upper layer is basically consistent. FM experiment appears a lower potential temperature value in the bottom layer than the initial moment. With the passage of time, the difference in the potential temperature profile is small at the 11th and 16th moments, and the two curves are basically coincident. It is shown that the development of the potential temperature of the cold bubble is basically consistent in the subsequent time

[0108] Figure 11The minimum potential temperatures for the cold bubble in the FM and WRF models differ significantly from those for the hot bubble. While the WRF minimum temperature rises continuously, the FM minimum temperature rises slowly from time 0 to 6, then drops sharply from time 6 to time 9 before rising sharply again. The maximum and minimum wind speeds simulated by the WRF and FM models differ slightly, occurring between time 6 and 12, with time 12 being the point of extreme value exchange. Overall, the differences are minimal.

[0109] Figure 12 As shown in the figure, the energy diagnosis of the cold bubble experiment is similar to the conclusion of the hot bubble experiment. The kinetic energy of the FM experiment and the WRF experiment is constantly increasing. After the cold bubble is grounded, the kinetic energy tends to be stable. The FM potential energy and the kinetic energy are almost in anti-phase, first decreasing and then tending to be stable, while the potential energy of the WRF experiment decreases less, the internal energy of the Fm experiment changes less, and the internal energy of the WRF cold bubble experiment changes more, exceeding the potential energy. From the perspective of total energy, the energy line of the Fm experiment tends to be flat, while the energy curve of the WRF experiment fluctuates greatly with the changing trend of internal energy and potential energy, and is ultimately dissipated.

[0110] From the perspective of energy conversion, the FM experiment and the WRF experiment are similar. Kinetic energy is converted from potential energy and internal energy. The difference is that the internal energy of the WRF experiment changes greatly.

[0111] This application starts with satisfying the energy properties of the model and, based on the energy properties of the continuous form of the non-hydrostatic primitive equations, constructs a non-hydrostatic model dynamic framework spatial discretization scheme that maintains the energy transmission and conversion characteristics of the local infinitesimal element. This scheme reconstructs the mutual conversion and constraint mechanism of energy components such as kinetic energy, gravitational potential energy, internal energy, and latent heat energy within the discrete infinitesimal element, as well as the energy distribution mechanism between each infinitesimal element. Through comparative experiments with the ideal experimental density flow case in WRF, the following results were obtained:

[0112] (1) Compared with the WRF hot bubble experiment, the FM temperature decay rate and amplitude are smaller during the hot bubble development process, which proves that the FM hot bubble has better energy fidelity.

[0113] (2) From the energy diagnosis of the hot bubble experiment, the kinetic energy changes of FM and WRF are relatively close. The potential energy of the FM experiment generally shows a trend of continuous attenuation, while the potential energy of the WRF experiment first rises and then stabilizes in the region. From the perspective of internal energy, the change amplitude of WRF is larger. Compared with the continuously decreasing total energy of WRF, the total energy of FM experiment is relatively stable. The FM experiment has better energy conservation properties.

[0114] (3) The results of the three cold-bubble experiments showed that the results of the cold-bubble experiments were more sensitive to changes in resolution than those of the WRF experiments.

[0115] (4) The energy diagnosis of the cold bubble experiment is similar to the hot bubble experiment. The change of FM potential energy is greater than WRF, while the change of internal energy is less than WRF, and the total energy change tends to be stable.

[0116] Overall, the FM density flow experiment has better overall conservation than the WRF experiment, and has better energy fidelity properties. The current case experiment is limited to density flow experiments, and more ideal experiments are needed to verify the conservation and stability of the model.

[0117] Mountain wave experiment

[0118] Compared with WRF, the differences between the two dynamic frameworks in the simulation process were explored through a two-dimensional mountain wave ideal experiment to test the simulation effect and conservation performance of the dynamic framework. The mountain wave ideal experiment is to simulate the propagation process of mountain waves caused by the undulation of terrain under the same initial field of uniform zonal wind field. The simulation parameter settings of the two experiments are shown in Table 2. In this ideal experiment, no microphysical scheme or radiation scheme is used.

[0119]

[0120]

[0121] Table 2. Model parameter settings.

[0122] The different grid spacings Δx selected for the two experiments are 200 m, 2 km and 20 km, respectively. The total number of grid points in the horizontal direction is 200, and the horizontal direction is periodic boundary. In the FM experiment, the vertical direction is equally spaced with Δz = 400 m, the model top height is 16 km, the absorption layer is at a height of 10 km, the initial field horizontal wind speed U = 10 m / s, N = 10 -2 s -1 , f = 10 -4 s -1 , and the clock-shaped mountain structure function used is:

[0123]

[0124] The mountain is located in the central region of the x-axis, and the mountain height h = 400 m. a is the horizontal width parameter of the mountain. According to the results of the influence of horizontal wavelength on amplitude obtained by JIMYDUDHIA (1991), in order to make the simulation process under 3 scales reach a steady-state solution, the mountain half-width parameter will change with the horizontal grid spacing of different resolutions a = 5Δx, and the total simulation time T = 21.6a / U, which corresponds to 36 min, 6 h and 60 h, respectively.

[0125] Figure 13The sensitivity experiments of the framework and WRF version at three scales. In Fig. a, Δx = 200m, the influence of gravity on the airflow over the mountain is most obvious because of the small time scale of the airflow, making the wave shape more inclined downstream, and the horizontal scale of the wave shape is close to the vertical scale. At the last moment of simulation, compared with the WRF version, the framework has a more detailed internal structure and propagates further to the leeward side of the wave. In Figs. b, c, Δx = 2km and 20km, the time scale of the airflow over the mountain is large, and the energy is obviously transmitted upwards, making the wave shape develop in the vertical direction. The wave shape in the WRF model tends to be smooth and uniform, while the wave shape in the fidelity framework is more detailed in the center of the extreme value. It can be seen that the propagation and fragmentation of the wave shape on both sides of the mountain are more detailed, and at the large scale of 20km resolution, the influence of the Coriolis force rotation makes the convergence and divergence centers tilt more obviously to the leeward side.

[0126] Energy spectrum characteristics

[0127] In the ideal experiment of two-dimensional mountain waves, the different energy conversions caused by terrain forcing are mainly kinetic energy, internal energy and potential energy. The above three groups of figures give the comparison of the time evolution of different energy perturbations under two versions, the left is the WRF version, and the right corresponds to the new design version, where all the energies are subtracted from the initial perturbation value (e T -e T0 Due to the influence of the absorbing layer, the total energy in space will gradually decrease with the simulation time, so the dispersion of the total energy in the figure is gradually increasing.

[0128] Figure 14 In the comparison figure at 200m resolution, due to the influence of terrain lifting on the airflow over the mountain, the air mass center rises as a whole, so both versions have obvious characteristics of internal energy to potential energy conversion at the initial few moments, and the kinetic energy gradually decreases with time. At the final moment of WRF simulation, the perturbation kinetic energy gradually decreases to-1.5×10 8And the changes of internal energy and potential energy tend to be smooth in the later half of the simulation; while in the new framework, the kinetic energy dispersion is relatively small, and the two periodic changes of internal energy and potential energy can be seen. At 2 km resolution, the energy development trends of the two frameworks are different, and the fidelity model has more detailed conversion characteristics than WRF, and the periodicity of the mutual conversion of internal energy and potential energy is more obvious. In the WRF version, the kinetic energy decays more severely. At 20 km resolution, the differences between the two frameworks are more significant, and the conversion process of internal energy and potential energy in the WRF version is obviously opposite to the previous two resolutions, and it is not yet clear what caused it, and the total energy dispersion is relatively more serious. It is worth mentioning that in the WRF version, the internal energy and potential energy both show sudden increases and decreases in the initial stage of the simulation, while in the new framework, the development of the two energies is relatively flat in the initial stage of the simulation, and the kinetic energy first increases and then decreases.

[0129] Overall, the simulation results in the new framework at three resolutions are relatively good, and the internal energy, potential energy, and kinetic energy develop relatively stably over time, making it easier to visually observe the periodic conversion process between potential energy and internal energy, and facilitating the correct description of the energy cascade process.

[0130] Vertical transport characteristics of horizontal momentum:

[0131] To intuitively reflect the performance of different dynamic frameworks in simulating nonlinear mountain waves, another reference element selected by the experiment is to test the vertical transport characteristics of horizontal momentum, so as to compare the sensitivity of different dynamic frameworks to the second-order effects of horizontal and vertical velocities. The vertical momentum flux of the mountain wave during propagation is defined as:

[0132]

[0133] According to the research of Miles and Huppert (1969) et al., under ideal conditions, given a mountain height and initial velocity, the vertical momentum flux of the mountain wave can be approximated as an analytical value constant:

[0134]

[0135] where ρ0 is the surface density, the Vaisala frequency N = 10 -2 s -1 , and the Scorer parameter l = 0.977 × 10 -3 m -1The closer the actual value M(Z) at different heights to the analytical value M in the simulation process, the closer the simulation effect under the framework to the actual situation. When the grid spacing Δx = 2000m, the group velocity and phase velocity of the mountain wave are close, the energy is vertically upward, the vertical wavelength is about one tenth of 2πa, and the horizontal wavelength is the strongest forced, so only the results of Δx = 2000m resolution and simulation total time T = 6h are analyzed.

[0136] Figure 15 The horizontal coordinate in the above figure is the ratio of the actual simulated vertical momentum flux to the theoretical value M(Z), and the vertical coordinate is selected as 0-6.4km (about one vertical wavelength). At the first time T = 1h, the vertical momentum fluxes of the two versions vary with height basically consistently, from close to 0 at high altitude to gradually converge to close to the theoretical value. As the simulation time goes on, the simulation results of the whole area will gradually tend to be stable. At the next five times, the simulated vertical momentum characteristics under the two frameworks are obviously different. In the WRF version, the simulation basically reaches a stable state at 2h, the calculation reaches a stable state in a relatively short time, and the curve characteristics of the vertical momentum basically coincide at the next few times. Overall, the momentum decays obviously with height, and the ratio of the actual momentum to the analytical value at the height of 6km at the final time is still less than 25%. In the right figure, it can be seen that after 5h of simulation under the new framework, the momentum changes begin to flatten, and the calculation process needs a relatively longer time to reach stability, but the simulation result is obviously closer to the analytical value. The momentum flux decays from the bottom to the highest layer and can be close to 60%, which is obviously better than the calculation result of the WRF version. Although more than 100% appears in the lowest layer in both versions, it is speculated that some constants used to solve the analytical value and the actual situation have some errors, so it is only used as an evaluation index of the simulation effect. Therefore, from the above experimental results, we can consider that in this ideal experiment of simulating the mountain wave, the discrete scheme under the new framework can obtain more accurate simulation results, and the calculated momentum flux is better matched with the analytical value.

[0137] As Figures 16-17 , a new fully elastic non-static energy preserving numerical scheme is constructed. In order to test and improve its simulation performance, this paper mainly compares with the WRF model, carries out ideal experiments of two-dimensional dry mountain wave under different resolutions, and draws the following conclusions:

[0138] (1) In the simulation comparison of three resolutions, the vertical velocity distribution of the two versions is slightly different below 10km. The calculation result of the WRF experiment is relatively smooth and uniform, and the results of the two experiments are close in numerical value, but the energy propagation performance of the mountain wave is better and the internal structure is more detailed from the vertical velocity profile of the FM experiment.

[0139] (2) From the results of the energy spectrum, the kinetic energy trends of the two experiments are gradually decreasing with the simulation time. The momentum equation in the form of Lamb is used in the process of constructing the kinetic energy equation in the FM experiment, which makes the calculation error of kinetic energy relatively better than that of the WRF experiment. From the process of mutual conversion of internal energy and potential energy, the periodic fluctuations of the two in the FM experiment are more detailed, the calculation of energy cascade process is more accurate, and the simulation results at different resolutions are also more stable.

[0140] (3) According to the vertical transport characteristics of horizontal momentum, the significant difference in calculation between the two frameworks can be intuitively examined. The characteristics of the WRF model are that it can reach calculation stability in a relatively short time, but the vertical momentum flux converges to the analytical value with height, and the effect becomes worse and worse, decaying more severely within a vertical wavelength. The FM experiment takes a relatively long time to reach calculation stability, but its momentum convergence performance is significantly better than that of the WRF experiment, basically exceeding 70% of the analytical value within a vertical wavelength, indicating that its energy propagation performance is better.

[0141] The present application relates to a new spatially discrete dynamic framework of a new spatially discrete scheme, which uses density, velocity and air pressure as forecast variables, adopts a new difference scheme and a momentum equation converted from Euler form to Gromicko-Lamb form. The spatially discrete format of the dynamic framework satisfies the mass conservation property, and the time discrete scheme adopts a semi-implicit time splitting algorithm. The momentum advection constructed by this can be expressed in the form of flux, and the motion equation retains the characteristics of effective energy transfer in the discrete format of the advection term, which can effectively reduce the error caused by kinetic energy advection calculation, thereby increasing the calculation accuracy of the mutual conversion of each energy.

[0142] It can be seamlessly coupled with any business system. Initial field data can be obtained from the Internet. The method performs modular design of process modules, has clear logical structure, is simple, efficient, has moderate calculation amount, occupies small storage space for data, has high calculation efficiency, is simple and easy to build system, is convenient for transplantation, and is easy to apply in meteorological business and various industry departments. At the same time, the historical forecast case database can be updated and expanded in real time, and new historical forecast cases are continuously integrated. The larger the sample capacity of historical forecast cases is, the more accurate the similar forecast member selection is, and the more ideal the forecast effect is.

[0143] This application has a wide range of applications. Meteorological business departments can use this numerical model to conduct numerical weather forecasts, providing scientific support for disaster reduction, agricultural production, and ensuring the safety of people's lives and property. It has great application potential. This technology has broad application prospects in the civil aviation sector. It accurately predicts the most important aviation-related dangerous weather for flight safety, including turbulence, turbulence, ice accumulation, lightning strikes (lightning strikes), downbursts, low-level wind shear and other weather phenomena that endanger flight safety. This forecasting method can improve the 72-hour forecast accuracy of airports and routes, and provide a reference for taking avoidance measures and planning routes in advance.

[0144] For operational meteorological departments, this forecasting method can seamlessly integrate with the numerical model forecasting systems of operational meteorological stations, providing more accurate weather forecasts for the nation and society. For the agricultural sector, it can support agricultural and food security. For transportation management departments, it can provide high wind and lightning forecasts, contributing to public transportation safety. For emergency and disaster relief departments, it can provide ensemble forecasts of severe weather events, providing scientific support for emergency measures. In summary, this method can meet the severe convective weather forecasting needs of various sectors, including agriculture, transportation, and disaster relief, and has broad application prospects and significant value.

[0145] The following will be combined with the Figure 18 , a method for constructing a dynamic framework of a numerical model for weather forecasting provided in an embodiment of the present application includes the following steps:

[0146] S1. Establish the basic equations of atmospheric motion and energy conversion relationships for the dynamic framework of the numerical model;

[0147] S2. Based on the basic equations of atmospheric motion of the numerical model dynamic framework and the energy conversion relationship, a dynamic framework spatial discrete model is established.

[0148] Furthermore, based on the above embodiment, the basic equations of atmospheric motion for establishing the numerical model dynamic framework specifically include:

[0149] The basic equations of atmospheric motion are the Lamb form of the NS equations

[0150]

[0151] Where V = (u, v, w) is the atmospheric wind field, K is the atmospheric kinetic energy K = |V 2 / 2, ρ is the atmospheric density, p is the atmospheric pressure, T is the atmospheric temperature, F represents the dissipative force, Q represents the heat source, q is the water vapor specific humidity, S represents the water vapor source and sink, Ω is the angular velocity of the Earth's rotation, g is the acceleration of gravity, c v is the atmospheric isochoric specific heat coefficient, and R is the atmospheric gas constant.

[0152] Further, based on the above embodiment, the energy conversion relationship of the atmospheric motion of the numerical model dynamic framework is established, specifically including:

[0153] The basic equation set of the atmospheric motion has the following energy conversion relationship

[0154]

[0155] Wherein p is the atmospheric density, E is the total energy, E = K + c v T + φ + Lq, c v T is the internal energy, φ is the gravitational potential energy Lq is the latent heat of water vapor;

[0156] When the system has no dissipation F = 0, and only considers the heat change Q = - LS caused by phase change, the energy conservation relationship of any point in space can be obtained,

[0157] Further, based on the above embodiment, further including:

[0158] The atmospheric motion equation set under the height terrain following vertical coordinate is established;

[0159] The horizontal coordinate of the atmospheric motion equation set is local coordinate (x, y), and the vertical coordinate is height terrain following coordinate Wherein z S = z S (x, y) is the ground height, z T is the top height of the atmosphere, which is taken as a constant;

[0160] In order to maintain the energy and mass properties, the basic prediction quantity adopts u, v, w, p, pT; wherein u is the zonal wind speed, v is the radial wind speed, w is the vertical speed, p is the atmospheric density, and pT is the product of density and temperature;

[0161] Map magnification factor m and n, here m, n are taken as equal, wherein m is the ratio of the grid distance in x direction in the map to the real grid distance in x direction, and n is the ratio of the grid distance in y direction in the map to the real grid distance in y direction;

[0162] The motion equation adopts non-flux non-advection form, and the specific form is as follows

[0163] Wherein

[0164]

[0165] Wherein r = a + z, a is the earth radius, is the latitude.

[0166] The equation set under the above terrain coordinate satisfies the following energy conversion relationship:

[0167]

[0168] where F x is the dissipation force in the x direction, F y is the dissipation force in the y direction, and F z is the dissipation force in the z direction.

[0169] Further, based on the above embodiment, the basic equations of atmospheric motion based on the numerical model dynamic framework and the energy conversion relationship, the dynamic framework space discrete model is established, specifically comprising:

[0170] The discrete grid distribution level of the forecast variable is Arakawa-C grid distribution, and the vertical is Lorenz distribution, and the horizontal uniform grid spacing Δx and Δy are adopted, and the vertical non-uniform grid spacing Δσ is adopted;

[0171] The following discrete operators are defined:

[0172]

[0173] The kinetic energy is defined on the Cross scalar point,

[0174] The most critical is to meet the kinetic energy flux form, that is, the momentum equation, that is, the wind field and the continuity equation, that is, the density related discrete scheme, meets the following kinetic energy flux relationship:

[0175]

[0176] Based on the above constraint condition, the following dynamic framework space discrete scheme is obtained:

[0177]

[0178] where δ is the height terrain following perturbation quantity, δ x is the x direction deviation, and δ y is the y direction deviation

[0179]

[0180] Further, based on the above embodiment, the basic equations of atmospheric motion based on the numerical model dynamic framework and the energy conversion relationship, the dynamic framework space discrete model is established, and further comprising:

[0181] The dynamic framework space discrete model in the form of perturbation is established:

[0182] The state quantity (ρ, T, p) is decomposed into basic quantity (ρ b , T b , pb ), plus the perturbation (ρ', T', p'), that is:

[0183] ρ=ρ b +ρ′,T=T b +T′,p=p b +p′

[0184] Assume that the basic quantity changes only with height and satisfies the static equilibrium relationship The basic temperature is given by the vertical profile of the mean temperature;

[0185] The atmospheric motion equations in the corresponding perturbation form are as follows:

[0186]

[0187] The corresponding discrete form is:

[0188] Furthermore, based on the above embodiment, the basic equations of atmospheric motion based on the numerical model dynamic framework and the energy conversion relationship are used to establish a dynamic framework spatial discrete model, which further includes:

[0189] Time discretization uses the classic semi-implicit HE-VI time splitting algorithm, where the fast wave correlation term is integrated with a semi-implicit small time step, and the rest is integrated with a large time step;

[0190] The following is a discrete model with a small time step, where Δτ is a small time step and Δt is a large time step;

[0191]

[0192] Among them, the vertical velocity of the underlying surface is:

[0193]

[0194] The following will be combined with the Figure 19 , an embodiment of the present application provides a device for constructing a numerical model dynamic framework for weather forecasting, the device comprising:

[0195] The first processing module is used to establish the basic equations of atmospheric motion and energy conversion relations of the numerical model dynamic framework;

[0196] The second processing module is used to establish a dynamic framework spatial discrete model based on the basic equations of atmospheric motion of the numerical model dynamic framework and the energy conversion relationship.

[0197] In addition, the embodiment of the present application comprises a computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the method for constructing a numerical model dynamic framework for weather forecasting according to any one of the above technical solutions when executing the computer program.

[0198] The embodiment of the present application also comprises a computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the method for constructing a numerical model dynamic framework for weather forecasting according to any one of the above technical solutions.

[0199] The above embodiments are only used to illustrate the technical solutions of the present application rather than limit the present application. Although the present application has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that the specific embodiments of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or equivalent replacement should be covered within the protection scope of the claims of the present application.

Claims

1. A numerical model dynamical framework construction method for weather prediction, characterized by, The method comprises: establishing a basic equation set of atmospheric motion and an energy conversion relationship of a numerical model dynamic framework; based on the basic equation set of atmospheric motion and the energy conversion relationship of the numerical model dynamic framework, a dynamic framework spatial discrete model is established; the establishment of the basic equation set of atmospheric motion of the numerical model dynamic framework specifically comprises: the basic equation set of atmospheric motion adopts N-S equation set in the form of Lamb where is the atmospheric wind field, K is the atmospheric kinetic energy , is the atmospheric density, is the atmospheric pressure, T is the atmospheric temperature, F represents the dissipative force, Q represents the heat source, q is the water vapor specific humidity, S represents the water vapor source and sink, Ω is the earth's rotation angular velocity, g is the gravitational acceleration, is the atmospheric specific heat at constant volume, R is the atmospheric gas constant; the establishment of the energy conversion relationship of atmospheric motion of the numerical model dynamic framework specifically comprises: ， where p is atmospheric density, E is total energy, , c v T is internal energy, is gravitational potential energy Lq is latent heat of water vapor; When the system is non-dissipative , and only the heat change caused by phase transition is considered , the energy conservation relation at any point in space is obtained ; The method further comprises: establishing an atmospheric motion equation group in a height-terrain-following vertical coordinate; horizontal coordinates of the atmospheric motion equation group adopt local coordinates (x, y), and a vertical coordinate adopts a height-terrain-following coordinate , wherein is a ground height, is a top height of an atmospheric layer, and is taken as a constant; in order to maintain energy and mass properties, the basic prediction quantity adopts u, v, w, p and pT; wherein u is a zonal wind speed, v is a radial wind speed, w is a vertical speed, p is an atmospheric density, and pT is a product of density and temperature; a map magnification factor m and n, here m and n are taken as equal, wherein m is a ratio of a grid distance in the x direction of the map to a real grid distance in the x direction, and n is a ratio of a grid distance in the y direction of the map to a real grid distance in the y direction; the motion equation adopts a non-flux non-advection form, and the specific form is as follows wherein wherein , is the earth radius, is the latitude; The equation group under the terrain coordinates above satisfies the energy conversion relationship as follows: wherein is the dissipation force in the x direction, is the dissipation force in the y direction, is the dissipation force in the z direction; The basic equation set of atmospheric motion based on the numerical mode dynamic framework and the energy conversion relationship formula establish a dynamic framework space discrete model, specifically including: the discrete grid distribution level of the forecast variable is Arakawa-C grid distribution, and the vertical is Lorenz distribution, and a uniform grid spacing is adopted , and a vertical non-uniform grid spacing ​ the following discrete operator is defined: , , , , , Kinetic energy is defined on the Cross scalar point, ; wherein the most critical is to meet the kinetic energy flux form, that is, to require that the momentum equation, that is, the wind field and the continuity equation, that is, the density related discrete scheme meet the following kinetic energy flux relationship: Based on the above constraints, the following power frame space discretization scheme is obtained: , wherein is a high terrain following disturbance, is an x-direction bias, is a y-direction bias 。 2. The method of claim 1, wherein, The basic equation group of atmospheric motion based on the numerical mode dynamic framework and the energy conversion relation formula, establishing a dynamic framework space discrete model, further comprises: establishing a perturbation form dynamic framework space discrete model: decomposing state quantity into basic quantity and perturbation quantity , namely: Assume that the basic quantities vary only with height and satisfy the static equilibrium relations The basic temperature is given by the mean temperature vertical profile; the corresponding perturbation form of the atmospheric motion equation is as follows: The corresponding discrete form is: 。 3. The method of claim 2, wherein, based on the basic equation set of atmospheric motion and the energy conversion relationship of the numerical model dynamic framework, the dynamic framework spatial discrete model further comprises: The following is a discrete model for small time steps, for small time steps, for large time steps; where the vertical velocity of the underlying surface is: 。 4. A numerical model dynamical framework building apparatus for weather prediction implementing the method of any one of claims 1 to 3, characterized in that, the time discrete adopts a classical semi-implicit HE-VI time splitting algorithm, a fast wave related term is integrated by using a semi-implicit small time step, and the remaining part is integrated by using a large time step; where is the atmospheric wind field, K is the atmospheric kinetic energy , is the atmospheric density, is the atmospheric pressure, is the atmospheric temperature, represents the dissipation force, represents the heat source, is the water vapor mixing ratio, represents the water vapor source / sink, is the Earth's rotation angular velocity, is the gravitational acceleration, is the atmospheric specific heat at constant volume, is the atmospheric gas constant; the device comprises: the first processing module is used for establishing a basic equation set of atmospheric motion and an energy conversion relationship of a numerical model dynamic framework ， wherein is the atmospheric density, is the total energy, , is the internal energy, is the gravitational potential energy , is the latent heat of water vapor; When the system is non-dissipative , and only the heat change due to phase transition is considered , the energy conservation relation at any point in space is obtained . 5.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-4 when the computer program is executed by the processor. the establishment of the energy conversion relationship of atmospheric motion of the numerical model dynamic framework specifically comprises:

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, the basic equation set of atmospheric motion has the following energy conversion relationship the processor executes the computer program to realize the numerical model dynamic framework construction method for weather prediction in any one of claims 1 to 3. the computer program is executed by the processor to realize the numerical model dynamic framework construction method for weather prediction in any one of claims 1 to 3.

Images