A cloud rain control variable layered adaptive gaussianization conversion method and device

Abstract

The application relates to the technical field of numerical weather prediction, in particular to a layered self-adaptive Gaussian conversion method and equipment for cloud and rain control variables, which can acquire observation data and input the observation data into a numerical prediction model to obtain three-dimensional cloud and rain control variables; a variational assimilation model including a forward transformation operator, an inverse transformation operator, a tangent linear operator and an adjoint operator is constructed; the forward transformation operator can calculate and adjust conversion strength layer by layer, realizing more fine and accurate Gaussian processing; the tangent linear operator and the adjoint operator are used for internal minimum iteration calculation of the assimilation model, and then the inverse transformation operator is used to restore the three-dimensional cloud and rain control variable analysis field to its original physical order of magnitude output, so that the actual three-dimensional cloud and rain control variable analysis field is obtained. According to the technical scheme, the vertical distribution difference of the cloud and rain variables can be accurately adapted, and through the provision of a complete operator chain, the compatibility with an existing assimilation system is ensured, so that the analysis field quality and the prediction ability are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of numerical weather prediction, in particular to a layered self-adaptive Gaussian conversion method and device of cloud and rain control variables. BACKGROUND

[0002] Extreme precipitation weather evolves dramatically, has small space-time scale and strong suddenness, and poses a severe challenge to existing numerical weather prediction systems. In recent years, research has shown that the reasonable introduction of cloud and rain control variables in the data assimilation process can effectively improve the simulation and prediction ability of strong precipitation systems. Cloud and rain variables carry key information such as water content, phase composition and microphysical conversion process, and their reasonable introduction helps to correct systematic biases caused by microphysical parameterization uncertainty in the initial field of the model, thereby generating an analysis field that is more consistent with the actual atmospheric state, laying a physical foundation for improving precipitation prediction accuracy.

[0003] The accuracy of numerical weather prediction is highly dependent on the quality of the initial field constructed by data assimilation. The variational assimilation method, as the core of the current operational assimilation system, is based on the basic assumption that the background error of the variable obeys the Gaussian distribution. This assumption ensures the convexity of the cost function of the variational assimilation method, which is a mathematical prerequisite for efficiently solving the optimal analysis field. However, this theoretical premise is being significantly challenged by strong nonlinear variables in high-resolution, multi-physical process models, with cloud and rain variables being typical representatives. Such variables often exhibit non-Gaussian properties such as skewness, truncation and asymmetric distribution in mesoscale and convective systems, directly undermining the optimality of the variational assimilation theory, and becoming a major bottleneck in further improving the prediction accuracy of extreme precipitation.

[0004] To solve this problem, some variable conversion methods have been proposed in existing technologies, aiming to map the non-Gaussian original variables to the control variable space with approximately Gaussian distribution. For example, a common method is to use a globally uniform nonlinear function (such as logarithmic transformation) to process cloud and rain variables. However, this method uses fixed function form and parameters for all vertical height layers, ignoring the significant heterogeneity of the statistical distribution of cloud and rain variables in the vertical structure, making it difficult to achieve ideal conversion results at all height layers, and easily causing numerical instability when the variable value is close to zero or extreme.

[0005] In addition, although some other transformation methods can improve the distribution characteristics of the variables to some extent, the transformation operators constructed by them often do not have strict reversibility or smoothness, making it impossible to derive the complete operator chain required by the variational assimilation minimization algorithm, especially the tangent linear operator and the adjoint operator. The irreversibility or gradient discontinuity of the operator will directly lead to the failure of the assimilation system in the minimization iterative solution process, which greatly limits the practical application value of these methods in operational variational assimilation systems. SUMMARY

[0006] Therefore, the present application aims to provide a layered adaptive Gaussianization conversion method and device for cloud and rain control variables, so as to solve the technical problem that the statistical characteristic difference of different spatial positions and the compatibility of the transformation operator chain and the operational assimilation system cannot be considered when the cloud and rain variables are Gaussianized in the prior art.

[0007] According to a first aspect of the present application, a layered adaptive Gaussianization conversion method for cloud and rain control variables is provided, comprising:

[0008] obtaining observation data and inputting the observation data into a numerical prediction model to obtain three-dimensional cloud and rain control variables;

[0009] constructing a variational assimilation model comprising a forward transformation operator, an inverse transformation operator, a tangent linear operator and an adjoint operator;

[0010] inputting the cloud and rain control variables into the variational assimilation model, calculating the standard deviation of the cloud and rain control variables at each height layer, inputting the standard deviation into the forward transformation operator, performing adaptive adjustment on the conversion intensity, and obtaining the cloud and rain control variables in the converted control variable space;

[0011] the variational assimilation model performs internal minimum iteration calculation on the converted cloud and rain control variables by using the tangent linear operator and the adjoint operator, and obtains new cloud and rain control variables in the control variable space;

[0012] the variational assimilation model restores the new cloud and rain control variables in the control variable space to their original physical magnitude by using the inverse transformation operator, and outputs an actual three-dimensional cloud and rain control variable analysis field.

[0013] Preferably, the forward transformation operator in the variational assimilation model is represented by the following formula:

[0014]

[0015] wherein, is the cloud and rain control variable in the physical space, is the converted cloud and rain control variable in the control variable space, is the standard deviation of the cloud and rain control variable in the physical space, is a vertical layer index, is a scaling operator.

[0016] Preferably, the inverse transformation operator in the variational assimilation model is represented by the following formula: .

[0017] Preferably, the tangent linear operator in the variational assimilation model is represented by the following formula:

[0018]

[0019] wherein, is a control variable space disturbance, is a corresponding disturbance in the physical space.

[0020] Preferably, the adjoint operator in the variational assimilation model is represented by the following formula: .

[0021] Preferably, the obtained observation data comprises:

[0022] sounding observation data, radar observation data and satellite observation data.

[0023] According to a second aspect of the present application, there is provided a layered adaptive Gaussianization conversion device for cloud-rain control variables, comprising:

[0024] a host controller, and a memory connected to the host controller;

[0025] the memory, wherein the memory has stored program instructions;

[0026] the host controller is configured to execute the program instructions stored in the memory to perform the method of any one of the above.

[0027] The technical solution provided by the present application can include the following beneficial effects:

[0028] It can be understood that the technical solution shown by the present application can obtain observation data and input the observation data into a numerical prediction model to obtain three-dimensional cloud-rain control variables; a variational assimilation model including a forward transformation operator, an inverse transformation operator, a tangent linear operator and an adjoint operator is constructed; the forward transformation operator can calculate and adjust the conversion intensity layer by layer to achieve more fine and accurate Gaussianization processing; the tangent linear operator and the adjoint operator are used for internal minimization iterative calculation of the assimilation model, and then the inverse transformation operator is used to restore the original physical order of magnitude output to conform to the actual three-dimensional cloud-rain control variable analysis field. The technical solution takes the three-dimensional cloud-rain control variables output by the numerical prediction model as the basic data, fully considers the three-dimensional structure and significant non-Gaussian characteristics of the cloud-rain control variables, and can adaptively adjust the conversion intensity at different height layers through the forward transformation operator to effectively improve the Gaussianity of the background error of each layer, thereby improving the calculation efficiency of the assimilation model and the quality of the analysis field obtained after assimilation. Moreover, by providing a complete operator chain, the compatibility with the existing assimilation system is ensured, thereby improving the analysis field quality and the prediction ability.

[0029] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. Attached Figure Description

[0030] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0031] Figure 1 This is a schematic diagram illustrating the steps of a hierarchical adaptive Gaussianization transformation method for data assimilation, according to an exemplary embodiment. Detailed Implementation

[0032] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0033] In one embodiment, Figure 1 This is a schematic diagram illustrating the steps of a hierarchical adaptive Gaussianization transformation method for data assimilation according to an exemplary embodiment. See also: Figure 1 This paper provides a hierarchical adaptive Gaussianization transformation method for cloud and rain control variables, including:

[0034] Step S11: Obtain observation data and input it into the numerical prediction model to obtain three-dimensional cloud and rain control variables.

[0035] In one embodiment, the acquired observation data includes: radiosonde observation data, radar observation data, and satellite observation data.

[0036] In practice, this step is the data input stage of the entire process. Specifically, a short-term forecast is made by running a numerical weather prediction model, and the three-dimensional cloud and rain control variables are obtained from the input observation data. Cloud and rain control variables may include, but are not limited to, water-water mixing ratio, rain-water mixing ratio, snow mixing ratio, and graupel mixing ratio, as well as other water condensate variables.

[0037] Step S12: Construct a variational assimilation model that includes forward transformation operator, inverse transformation operator, tangent linear operator and adjoint operator.

[0038] The variational assimilation model has a variational assimilation framework that fully integrates the forward transformation operator, inverse transformation operator, tangent linear operator, and adjoint operator into the cloud and rain control variable quality control module of the variational assimilation framework.

[0039] The forward transform operator transforms the cloud-rain control variable so that its background error is more Gaussian. In an embodiment, the forward transform operator is represented by the following equation:

[0040]

[0041] wherein, is the cloud-rain control variable in the physical space, is the transformed cloud-rain control variable in the control variable space, is the standard deviation of the cloud-rain control variable in the physical space, is the vertical layer index, which indicates that the transformation is performed layer by layer, i.e., each layer has its own statistical characteristics and transformation parameters, achieving the effect of layer-adaptive transformation. is the scaling operator, which is used to adjust the influence degree of the standard deviation on the overall scaling, and is set to 100 based on experience.

[0042] The inverse transform operator is the strict inverse process of the forward transform formula, which can accurately transform the cloud-rain variable in the control variable space back to the physical space after the assimilation iteration process is completed, to restore the magnitude of the actual cloud-rain variable. In an embodiment, the inverse transform operator in the variational assimilation model is represented by the following equation: .

[0043] The tangent linear operator and the adjoint operator are closely related and are the core algorithm of variational assimilation, i.e., the formula necessary in the minimization process.

[0044] The tangent linear operator is an approximation of the Jacobian matrix of the inverse transform operator at . Under the linearization assumption, it describes how a small perturbation in the control variable space is linearly mapped to a perturbation in the physical space. In an embodiment, the tangent linear operator in the variational assimilation model is represented by the following equation:

[0045]

[0046] wherein, is the perturbation in the control variable space, is the corresponding perturbation in the physical space.

[0047] The adjoint operator is the transpose of the tangent linear operator. In variational assimilation, it is used to back-propagate the gradient in the physical space to the control variable space. The adjoint operator in the variational assimilation model is represented by the following equation: .

[0048] ​By integrating the four operators into the variational assimilation model, adaptive optimization processing of the statistical characteristics of cloud and rain variables at different height layers is realized. At the same time, the complete operator chain ensures that the transformation method can be seamlessly embedded into all variational assimilation frameworks.

[0049] In step S13, the cloud and rain control variables are taken as inputs of the variational assimilation model, the standard deviations of the cloud and rain control variables at each height layer are calculated, the standard deviations are introduced into the forward transformation operator, the conversion strength is adaptively adjusted, and the cloud and rain control variables in the converted control variable space are obtained.

[0050] In specific practice, the assimilation model is run, the standard deviations of the cloud and rain variables at each height layer are calculated in the forward transformation operator, and the standard deviations at each height layer are introduced into the conversion formula of the forward transformation operator to realize adaptive adjustment of the conversion strength.

[0051] In step S14, the variational assimilation model uses the tangent linear operator and the adjoint operator to perform internal minimization iteration calculation on the converted cloud and rain control variables to obtain new cloud and rain control variables in the control variable space.

[0052] In step S15, the variational assimilation model uses the inverse transformation operator to restore the new cloud and rain control variables in the control variable space to their original physical magnitude, and outputs a three-dimensional cloud and rain control variable analysis field conforming to reality.

[0053] In this step, after the assimilation iteration converges, the inverse transformation operator integrated in the assimilation model is used to restore the newly obtained cloud and rain control variables to their original physical magnitude, and finally a high-quality three-dimensional cloud and rain control variable analysis field conforming to reality is output, providing a more ideal data basis for subsequent numerical prediction and business application.

[0054] It can be understood that the technical solution shown in the present application realizes layered adaptive Gaussianization, and the present application breaks through the limitations of traditional single global transformation and can effectively capture and adapt to the characteristics of cloud and rain variable distribution at different vertical heights. By layer-by-layer calculation and adjustment of the conversion strength, more fine and accurate Gaussianization processing is realized, so that the background error of each layer of cloud and rain variables is significantly improved, laying a solid foundation for generating a higher-quality analysis field.

[0055] The numerical values of the cloud and rain control variables before and after conversion remain continuous: the conversion method adopted by the present application does not use segmentation processing, ensuring the continuity of the numerical values of the cloud and rain variables before and after conversion, effectively avoiding the situation that unreasonable physical characteristics appear in the analysis field obtained by assimilation due to non-continuous numerical values.

[0056] The cloud and rain control variable background error is improved, and the operation efficiency of the assimilation model is significantly improved. Through actual case testing, after the method is applied, the quality of the analysis field obtained by assimilation is significantly improved.

[0057] The method has assimilation model compatibility and practicality. The core requirements of the variational assimilation system are considered. The forward transformation operator not only transforms effectively, but also provides complete, stable and computationally efficient inverse transformation operators, tangent linear operators and adjoint operators, ensuring that the method can be directly embedded into an existing operational variational assimilation system, meeting all mathematical and computational requirements, and having great potential for business application.

[0058] In another embodiment, a cloud and rain control variable layered adaptive Gaussian conversion device is provided, comprising:

[0059] A host controller and a memory connected to the host controller;

[0060] The memory, wherein the memory stores program instructions;

[0061] The host controller is configured to execute the program instructions stored in the memory to perform the method of any one of the above embodiments.

[0062] It can be understood that the same or similar parts in the above embodiments can be mutually referenced, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0063] It should be noted that in the description of the present application, the terms "first", "second" and the like are only used for description purposes and should not be understood as indicating or implying relative importance. In addition, in the description of the present application, unless otherwise specified, "a plurality of" means at least two.

[0064] Any process or method descriptions in flow charts or otherwise described herein can be understood as representing code modules, segments, or portions of code which include one or more executable instructions for performing specified logic functions or steps, and the various embodiments of the present application include additional implementations in which the functions described are performed by a state machine, or in which the functions are performed by circuits or hardware components that are external to a general purpose or special purpose computer, and that are programmed (either at the time of manufacture or at some time in response to a command), to perform specified tasks.

[0065] It should be understood that each part of the present application can be realized by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be realized by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if realized by hardware, and as in another embodiment, it can be realized by any one or a combination of the following technologies known in the art: discrete logic circuit with logic gate circuit for implementing logic function on data signal, application specific integrated circuit with suitable combination logic gate circuit, programmable gate array (PGA), field programmable gate array (FPGA) and the like.

[0066] Those skilled in the art of the present technology can understand that all or part of the steps carried out by the above-mentioned embodiment method can be completed by a program instructing the relevant hardware, and the program can be stored in a computer readable storage medium, and when executed, includes one or a combination of steps of the embodiment method.

[0067] In addition, each functional unit in each embodiment of the present application can be integrated in one processing module, or each unit can exist physically, or two or more units can be integrated in one module. The above-mentioned integrated module can be realized in the form of hardware or in the form of software functional module. The integrated module, if realized in the form of software functional module and sold or used as an independent product, can also be stored in a computer readable storage medium.

[0068] The above-mentioned storage medium can be read-only memory, disk or optical disk, etc.

[0069] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above-mentioned terms does not necessarily mean the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0070] Although the embodiments of the present application have been shown and described above, it should be understood that the above-mentioned embodiments are exemplary and cannot be understood as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above-mentioned embodiments within the scope of the present application.

Claims

1. A layered adaptive Gaussianization conversion method of cloud rain control variable, characterized in that, The method comprises: obtaining observation data and inputting the observation data into a numerical prediction model to obtain three-dimensional cloud and rain control variables; constructing a variational assimilation model comprising a forward transformation operator, an inverse transformation operator, a tangent linear operator and an adjoint operator; inputting the cloud and rain control variables into the variational assimilation model as inputs, calculating standard deviations of the cloud and rain control variables at each height layer, inputting the standard deviations into the forward transformation operator, performing adaptive adjustment and conversion of the conversion strength, and obtaining cloud and rain control variables in the converted control variable space; the variational assimilation model performs internal minimization iteration calculation on the converted cloud and rain control variables by using the tangent linear operator and the adjoint operator, and obtains new cloud and rain control variables in the control variable space; the variational assimilation model restores the new cloud and rain control variables in the control variable space to their original physical quantity level by using the inverse transformation operator, and outputs a three-dimensional cloud and rain control variable analysis field conforming to reality; the forward transformation operator in the variational assimilation model is represented by the following formula: wherein, is the cloud rain control variable in the physical space, is the transformed cloud rain control variable in the control variable space, is the standard deviation of the cloud rain control variable in the physical space, is the vertical layer index, is the scaling operator; the inverse transformation operator in the variational assimilation model is represented by the following formula: the tangent linear operator in the variational assimilation model is represented by the following formula: wherein, is a control variable space, is a corresponding disturbance in the physical space; the adjoint operator in the variational assimilation model is represented by the following formula: 。 2. The method of claim 1, wherein, the obtained observation data comprises: sounding observation data, radar observation data and satellite observation data.

3. A layered adaptive Gaussianization conversion device for cloud-rain control variable, characterized in that, The method comprises: a host controller and a memory connected to the host controller; the memory, wherein the memory stores program instructions; the host controller is configured to execute the program instructions stored in the memory to perform the method according to any one of claims 1-2.

Images