A live analysis method, system, medium, and device based on dynamic downscaling

Abstract

The present application relates to the technical field of meteorological analysis, more particularly, to a real-time analysis method, system, medium and equipment based on dynamic downscaling. The scheme includes cold start global model, parameter extraction, obtaining initial field and boundary field at current time as background field, hot start mesoscale model, generating high resolution forecast field, 1 kilometer resolution hot start interval, forming mesoscale assimilation observation data, and assimilation based on high resolution forecast field, generating mesoscale forecast field, according to small scale parameters, multi-scale fusion judgment is carried out, and analysis and prediction data under different scales are formed, according to high resolution forecast field, mesoscale forecast field and analysis and prediction data, superposition display is carried out, and multi-scale nested display data is formed. The scheme can make real-time analysis field, use high resolution model for short-term simulation, assimilate small scale observation as much as possible, and provide better quality background field for real-time analysis system.

Description

Technical Field

[0001] This invention relates to the field of meteorological analysis technology, and more specifically, to a real-time analysis method, system, medium, and equipment based on dynamic downscaling. Background Technology

[0002] Severe convective weather, hail, tornadoes, and other hazardous weather events pose significant challenges to current short-term weather forecasting. Accurate real-time analysis is essential for refined forecasting. With the development of meteorological observation systems, the density of observation station networks has increased year by year. For example, in eastern China, the number of automatic ground stations has reached a spatial range of 5 kilometers. In three-dimensional space, the construction of wind profiler radar, lidar, microwave radiometers, and the Fengyun-4 series satellites has effectively filled observational gaps. However, areas with insufficient observation still exist, particularly in three-dimensional spatial detection.

[0003] Prior to this invention, existing real-time analysis systems used forecast fields from mainstream operational models as background fields and employed data fusion techniques to create real-time analysis fields. However, these systems suffered from low spatial resolution, and the model's data assimilation system, constrained by model equilibrium assumptions, could not fully absorb small- and medium-scale observations. Summary of the Invention

[0004] In view of the above problems, this invention proposes a real-world analysis method, system, medium and equipment based on dynamic downscaling. The real-world analysis field is generated by a numerical simulation downscaling method, and a high-resolution model is used for short-term simulation to assimilate small- and medium-scale observations, especially radar observations, as much as possible, thereby providing a better quality background field for the real-world analysis system.

[0005] According to a first aspect of the present invention, a real-world analysis method based on dynamic downscaling is provided.

[0006] In one or more embodiments, preferably, the field analysis method based on dynamic downscaling includes:

[0007] The global model is cold-started to extract parameters and obtain the initial field and boundary field at the current moment, which are used as the background field.

[0008] Based on the background field, a hot-start mesoscale model is used to generate a high-resolution forecast field;

[0009] A 1-kilometer resolution hot start interval is used to form mesoscale assimilated observation data, which is then assimilated based on the high-resolution forecast field to generate a mesoscale forecast field.

[0010] Small-scale analysis is performed according to a preset 1-hour cycle to obtain small-scale parameters;

[0011] Based on the small-scale parameters, multi-scale fusion judgment is performed to form analysis and forecast data at different scales;

[0012] The high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data are overlaid and displayed to form a multi-scale nested display data.

[0013] In one or more embodiments, preferably, the cold-start global mode performs parameter extraction to obtain the initial field and boundary field at the current moment as the background field, specifically including:

[0014] At 2:00 AM Beijing time, global model analysis was initiated, along with initial forecast data.

[0015] The initial forecast data, with a resolution of 9 kilometers, is used to extract the initial field and the horizontal lateral boundary field as the background field.

[0016] In one or more embodiments, preferably, generating a high-resolution forecast field based on the background field and a hot-start mesoscale model specifically includes:

[0017] Based on the background field, at background time 3, the hot start mode is activated;

[0018] Upon receiving the hot start mode, a hot start mesoscale mode command is issued, and the changes in the current background field are automatically analyzed. The background field at the current moment is used as the forecast field of the previous moment and stored to generate a high-resolution forecast field.

[0019] In one or more embodiments, preferably, the step of performing a 1-kilometer resolution hot start interval to form mesoscale assimilated observation data, and then assimilating it based on the high-resolution forecast field to generate a mesoscale forecast field, specifically includes:

[0020] Upon receiving the hot-start mesoscale mode command, it performs 3DVar-based data acquisition at a resolution of 1 km to generate radar data.

[0021] Radar data is directly assimilated to form a synchronous field with complex cloud and water materials, which serves as a mesoscale forecast field.

[0022] In one or more embodiments, preferably, the step of performing small-scale analysis according to a preset 1-hour cycle to obtain small-scale parameters specifically includes:

[0023] Small-scale analysis is performed according to a preset 1-hour cycle;

[0024] Obtain all radar data, satellite data, and geographic information to obtain small-scale parameters.

[0025] In one or more embodiments, preferably, the step of performing multi-scale fusion judgment based on the small-scale parameters to form analysis and forecast data at different scales specifically includes:

[0026] Based on the small-scale parameters, each parameter is updated once, and the update time is recorded as the single data update time.

[0027] The scale classification of each parameter is calculated using the first calculation formula;

[0028] The long-term importance of each parameter is calculated using the second calculation formula;

[0029] The optimal target prediction count and the optimal target prediction duration are extracted using the third, fourth, and fifth calculation formulas. The calculation is performed when the scale level corresponding to the values ​​of the third and fourth calculation formulas is 1 or 0. When the scale level is 2, the optimal target prediction count for the corresponding parameter is set to a preset fixed value, and the optimal target prediction duration for the corresponding parameter is the product of the second level margin and the preset fixed value.

[0030] Based on the number of predictions for the optimal target and the prediction duration for the optimal target, analytical forecast data at different scales are generated.

[0031] The first calculation formula is:

[0032]

[0033] Wherein, C represents the scale classification, Y0 represents the first classification margin, Y1 represents the second classification margin, and D represents the single data update time.

[0034] The second calculation formula is:

[0035]

[0036] Where Z represents the importance of the long duration, T is the preset fluctuation analysis period, t is time, ΔB is the fluctuation of the parameter per unit time, and B AVG The mean of the parameters;

[0037] The third calculation formula is:

[0038] Y = PJ C

[0039] Where Y is the prediction accuracy index, J is the preset iteration accuracy, and P is the number of predictions per unit time;

[0040] The fourth calculation formula is:

[0041] A = Y / SU p

[0042] Where A is the predicted balance, SU p This represents the computing power required when the number of predictions per unit time is P.

[0043] The fifth calculation formula is:

[0044]

[0045] Among them, A i Let Z be the predicted balance of the i-th parameter, Z be the long-term importance of the i-th parameter, and P be the predicted balance. i T represents the number of times the optimal target is predicted for the i-th parameter. Y_i Let SU be the prediction duration of the optimal objective for the i-th parameter. i P represents the computation time complexity of the i-th parameter per unit time. i S represents the number of predictions per unit time for the i-th parameter. ALL N represents the total computing power per unit time, and N represents the total number of parameters.

[0046] In one or more embodiments, preferably, the step of overlaying and displaying the high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data to form multi-scale nested display data specifically includes:

[0047] According to a preset period, the high-resolution forecast field, the mesoscale forecast field, and the analysis and forecast data are extracted;

[0048] The high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data are assimilated to form an assimilated forecast field.

[0049] The data of the assimilation forecast field is displayed on a single screen.

[0050] According to a second aspect of the present invention, a real-time analysis system based on dynamic downscaling is provided.

[0051] In one or more embodiments, preferably, the real-world analysis system based on dynamic downscaling includes:

[0052] The global model module is used to cold start the global model, extract parameters, and obtain the initial field and boundary field at the current moment as the background field.

[0053] A cold-hot start module is used to hot-start a mesoscale model based on the background field to generate a high-resolution forecast field;

[0054] The mesoscale forecast module is used to perform a 1-kilometer resolution hot start interval to form mesoscale assimilated observation data, and then assimilate it based on the high-resolution forecast field to generate the mesoscale forecast field.

[0055] The small-scale assimilation module is used to perform small-scale analysis according to a preset 1-hour cycle to obtain small-scale parameters.

[0056] The multi-scale fusion module is used to perform multi-scale fusion judgment based on the small-scale parameters to form analysis and forecast data at different scales;

[0057] The rendering and display module is used to overlay and display the high-resolution forecast field, the mesoscale forecast field, and the analysis and forecast data to form multi-scale nested display data.

[0058] According to a third aspect of the present invention, a computer-readable storage medium is provided that stores computer program instructions thereon, which, when executed by a processor, implement the method as described in any one of the first aspects of the present invention.

[0059] According to a fourth aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method described in any one aspect of the present invention.

[0060] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0061] This invention's approach uses numerical simulation to compensate for the insufficient spatiotemporal resolution of data from small- and medium-scale weather systems, reconstructing near-surface three-dimensional wind and thermal fields. This is an important tool for studying the dynamics of small- and medium-scale weather systems. Small-scale weather systems like tornadoes, limited by mesoscale atmospheric models and computer capabilities, are primarily studied and used for historical case simulations. Currently, operational short-term forecasting models can achieve a regional 3-kilometer level.

[0062] This invention primarily aims to create real-time analysis fields without requiring long-term forecasts. Therefore, it incorporates small- to medium-scale observations into the model's initial field, enabling model integration to generate small- to medium-scale systems. This allows the model to force out small-scale weather systems with limited observations, providing a background field containing a relatively complete weather system in the real-time analysis. Furthermore, a multi-scale assimilation method is employed to fuse multi-source observation data, particularly radar observations, to improve the three-dimensional thermodynamic structure.

[0063] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0064] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0066] Figure 1 This is a flowchart of a real-world analysis method based on dynamic downscaling, according to an embodiment of the present invention.

[0067] Figure 2 This is a flowchart of a real-world analysis method based on dynamic downscaling, which is a cold-start global model in one embodiment of the present invention. Parameters are extracted to obtain the initial field and boundary field at the current moment, which serve as the background field.

[0068] Figure 3 This is a flowchart illustrating a real-world analysis method based on dynamic downscaling, according to an embodiment of the present invention, for generating a high-resolution forecast field by hot-starting a mesoscale model based on the background field.

[0069] Figure 4 This is a flowchart of a real-world analysis method based on dynamic downscaling in one embodiment of the present invention, which involves performing a 1-kilometer resolution hot start interval to form mesoscale assimilated observation data, and then assimilating it based on the high-resolution forecast field to generate a mesoscale forecast field.

[0070] Figure 5 This is a flowchart illustrating a dynamic downscaling-based real-world analysis method according to an embodiment of the present invention, which involves performing small-scale analysis according to a preset 1-hour cycle to obtain small-scale parameters.

[0071] Figure 6 This is a flowchart illustrating a real-world analysis method based on dynamic downscaling according to an embodiment of the present invention, which involves multi-scale fusion judgment based on the small-scale parameters to form analysis and forecast data at different scales.

[0072] Figure 7 This is a flowchart illustrating a dynamic downscaling-based real-time analysis method according to an embodiment of the present invention, which involves overlaying and displaying the high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data to form a multi-scale nested data display.

[0073] Figure 8 This is a structural diagram of a real-world analysis system based on dynamic downscaling, according to an embodiment of the present invention.

[0074] Figure 9This is a structural diagram of an electronic device according to one embodiment of the present invention.

[0075] Figure 10 This is a schematic diagram of the process and timeline for downscaling nested forecasting and analysis.

[0076] Figure 11 This is a schematic diagram of the results based on dynamic downscaling analysis. Detailed Implementation

[0077] In some of the processes described in the specification, claims, and accompanying drawings of this invention, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0078] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0079] Severe convective weather, hail, tornadoes, and other hazardous weather events pose significant challenges to current short-term weather forecasting. Accurate real-time analysis is essential for refined forecasting. With the development of meteorological observation systems, the density of observation station networks has increased year by year. For example, in eastern China, the number of automatic ground stations has reached a spatial range of 5 kilometers. In three-dimensional space, the construction of wind profiler radar, lidar, microwave radiometers, and the Fengyun-4 series satellites has effectively filled observational gaps. However, areas with insufficient observation still exist, particularly in three-dimensional spatial detection.

[0080] Prior to this invention, existing real-time analysis systems used forecast fields from mainstream operational models as background fields and employed data fusion techniques to create real-time analysis fields. However, these systems suffered from low spatial resolution, and the model's data assimilation system, constrained by model equilibrium assumptions, could not fully absorb small- and medium-scale observations.

[0081] As resolution increases, model terrain resolution also improves. However, due to limitations in the computational methods of the model's dynamic framework, high-resolution simulations can easily lead to instability after assimilating observations from radar and other sources. Therefore, regional operational models consider longer-term forecasts and filter out small- and medium-scale information. The data assimilation system avoids assimilating or sparses observations, or assimilates less small- and medium-scale observation information to maintain the model's integral stability.

[0082] This invention provides a method, system, medium, and device for real-world analysis based on dynamic downscaling. This approach generates a real-world analysis field using a numerical simulation downscaling method, employs a high-resolution model for short-term simulations, and assimilates small- and medium-scale observations, particularly radar observations, as much as possible, thereby providing a higher-quality background field for the real-world analysis system.

[0083] According to a first aspect of the present invention, a real-world analysis method based on dynamic downscaling is provided.

[0084] Figure 1 This is a flowchart of a real-world analysis method based on dynamic downscaling, according to an embodiment of the present invention.

[0085] In one or more embodiments, preferably, the field analysis method based on dynamic downscaling includes:

[0086] S101, Cold start global mode, extract parameters to obtain the initial field and boundary field at the current moment, as the background field;

[0087] S102. Based on the background field, a hot-start mesoscale model is used to generate a high-resolution forecast field;

[0088] S103. Perform a 1-kilometer resolution hot start interval to form mesoscale assimilation observation data, and assimilate it based on the high-resolution forecast field to generate a mesoscale forecast field.

[0089] S104. Perform small-scale analysis according to the preset 1-hour cycle to obtain small-scale parameters;

[0090] S105. Based on the small-scale parameters, perform multi-scale fusion judgment to form analysis and forecast data at different scales;

[0091] S106. The high-resolution forecast field, the mesoscale forecast field, and the analysis and forecast data are overlaid and displayed to form multi-scale nested display data.

[0092] In this embodiment of the invention, the system is designed primarily to create a real-time analysis field. Based on the determined background field, through analysis and overlay at different time scales, it introduces small-scale automatic control based on computing power and the current importance of the scale. This allows for free and automatic control while fulfilling the multi-scale premise, without requiring long-term forecasts. The model integration generates small- and medium-scale systems, enabling the model to force out small-scale weather systems with a limited number of observations. This provides a background field containing a relatively complete weather system in the real-time analysis. Furthermore, a multi-scale assimilation method is used to fuse multi-source observation data and improve the three-dimensional thermodynamic structure.

[0093] Figure 2 This is a flowchart of a real-world analysis method based on dynamic downscaling, which is a cold-start global model in one embodiment of the present invention. Parameters are extracted to obtain the initial field and boundary field at the current moment, which serve as the background field.

[0094] like Figure 2 As shown, in one or more embodiments, preferably, the cold-start global mode performs parameter extraction to obtain the initial field and boundary field at the current moment as the background field, specifically including:

[0095] S201, at 2:00 AM Beijing time, initiate global model analysis and initial forecast data;

[0096] S202. The initial forecast data, with a resolution of 9 kilometers, is used to extract the initial field and the horizontal side boundary field as the background field.

[0097] In this embodiment of the invention, at 02:00 Beijing time (18:00 UTC), the global model forecast field serves as the initial field and horizontal lateral boundary field of the mesoscale model. A cold-start 9-kilometer resolution mesoscale model's forecast field serves as the higher-resolution background field.

[0098] Figure 3 This is a flowchart illustrating a real-world analysis method based on dynamic downscaling, according to an embodiment of the present invention, for generating a high-resolution forecast field by hot-starting a mesoscale model based on the background field.

[0099] like Figure 3 As shown, in one or more embodiments, preferably, the step of generating a high-resolution forecast field by hot-starting a mesoscale model based on the background field specifically includes:

[0100] S301. Based on the background field, start the hot start mode when the background time is 3.

[0101] S302. Upon receiving the hot start mode, issue a hot start mesoscale mode command and automatically analyze the changes in the current background field, using the current background field as the forecast field of the previous time step and storing it to generate a high-resolution forecast field.

[0102] In this embodiment of the invention, model cold start refers to the use of a regional model's background field provided by other low-resolution models (such as global models). This background field is then assimilated or used as the initial field for forecasting. This ensures that the model forecast does not deviate from the large-scale background or environmental field. However, since the initial field lacks mesoscale information and generally does not contain cloud water content, the model requires a period of spin-up. 02:00 is chosen as the cold start time daily, primarily to ensure that the regional model's forecast does not deviate from the large-scale background or environmental field. In East Asia, 02:00 is nighttime, the boundary layer is relatively stable, and convective activity is inactive.

[0103] Figure 4 This is a flowchart of a real-world analysis method based on dynamic downscaling in one embodiment of the present invention, which involves performing a 1-kilometer resolution hot start interval to form mesoscale assimilated observation data, and then assimilating it based on the high-resolution forecast field to generate a mesoscale forecast field.

[0104] like Figure 4 As shown, in one or more embodiments, preferably, the step of performing a 1-kilometer resolution hot start interval to form mesoscale assimilated observation data, and then assimilating it based on the high-resolution forecast field to generate a mesoscale forecast field, specifically includes:

[0105] S401. Upon receiving the hot-start mesoscale mode command, perform 3DVar-based data acquisition at a resolution of 1 km to generate radar data.

[0106] S402. Radar data is directly assimilated to form a synchronous field with complex cloud and water substances, which serves as a mesoscale forecast field.

[0107] In this embodiment of the invention, during model hot start-up, the background field is assimilated from the forecast field of the previous time step. This forecast field already contains cloud and water content. During assimilation, the cloud and water content is directly assimilated using radar data or adjusted using complex cloud analysis. Assimilating radar radial velocity to directly adjust the wind field can effectively force the environmental conditions for the occurrence and development of small- and medium-scale weather systems. At the same time, assimilating radar reflectivity and radial velocity can comprehensively adjust the wind field, temperature field, and humidity field, so that the initial field achieves coordination in terms of dynamics, thermodynamics, and water vapor.

[0108] Figure 5 This is a flowchart illustrating a dynamic downscaling-based real-world analysis method according to an embodiment of the present invention, which involves performing small-scale analysis according to a preset 1-hour cycle to obtain small-scale parameters.

[0109] like Figure 5 As shown, in one or more embodiments, preferably, the step of performing small-scale analysis according to a preset 1-hour cycle to obtain small-scale parameters specifically includes:

[0110] S501. Perform small-scale analysis according to the preset 1-hour cycle;

[0111] S502: Obtain all radar data, satellite data, and geographic information to obtain small-scale parameters.

[0112] In this embodiment of the invention, the current mainstream model data assimilation system is based on Bayesian theory, statistically obtaining the background field error covariance and observation error covariance of the model, and using these as weights to minimize the analysis field and make predictions. Therefore, although some observations are accurate, the statistically obtained weights cannot completely approximate the observations. This invention, however, achieves multi-scale fusion of information through sequential changes at multiple scales, and considers multiple time layers to ensure the temporal continuity of the analysis.

[0113] Figure 6 This is a flowchart illustrating a real-world analysis method based on dynamic downscaling according to an embodiment of the present invention, which involves multi-scale fusion judgment based on the small-scale parameters to form analysis and forecast data at different scales.

[0114] like Figure 6 As shown, in one or more embodiments, preferably, the step of performing multi-scale fusion judgment based on the small-scale parameters to form analysis and forecast data at different scales specifically includes:

[0115] S601. Based on the small-scale parameters, update the data for each parameter once, and record the update time as the single data update time;

[0116] S602. Calculate the scale classification of each parameter using the first calculation formula;

[0117] S603. Calculate the long-term importance of each parameter using the second calculation formula;

[0118] S604. Extract the optimal target prediction count and the optimal target prediction duration using the third, fourth, and fifth calculation formulas together. The calculation is performed when the scale level corresponding to the values ​​of the third and fourth calculation formulas is 1 or 0. When the scale level is 2, the optimal target prediction count for the corresponding parameter is set to a preset fixed value, and the optimal target prediction duration for the corresponding parameter is the product of the second level margin and the preset fixed value.

[0119] S605. Based on the number of predictions for the optimal target and the prediction duration for the optimal target, generate analysis and forecast data at different scales;

[0120] The first calculation formula is:

[0121]

[0122] Wherein, C represents the scale classification, Y0 represents the first classification margin, Y1 represents the second classification margin, and D represents the single data update time.

[0123] The second calculation formula is:

[0124]

[0125] Where Z represents the importance of the long duration, T is the preset fluctuation analysis period, t is time, ΔB is the fluctuation of the parameter per unit time, and B AVG The mean of the parameters;

[0126] The third calculation formula is:

[0127] Y = PJ C

[0128] Where Y is the prediction accuracy index, J is the preset iteration accuracy, and P is the number of predictions per unit time;

[0129] The fourth calculation formula is:

[0130] A = Y / SU p

[0131] Where A is the predicted balance, SU p This represents the computing power required when the number of predictions per unit time is P.

[0132] The fifth calculation formula is:

[0133]

[0134] Among them, A i Let Z be the predicted balance of the i-th parameter, Z be the long-term importance of the i-th parameter, and P be the predicted balance. i T represents the number of times the optimal target is predicted for the i-th parameter. Y_i Let SU be the prediction duration of the optimal objective for the i-th parameter. i P represents the computation time complexity of the i-th parameter per unit time. i S represents the number of predictions per unit time for the i-th parameter. ALL N represents the total computing power per unit time, and N represents the total number of parameters.

[0135] In this embodiment of the invention, the optimal measurement parameters are set by combining the first, second, third, fourth, and fifth calculation formulas to achieve a balance between the number of predictions and the total prediction time. This allows for scale-level analysis based on different dynamic downscaling methods without overflowing computing power. Each analysis begins at a preset time interval. The main purpose of the analysis is to determine the time required to update the data for each small-scale observation within the current segment. The scale level is determined based on the time duration: C = 0 represents a microscale (fastest computation), C = 1 represents a fine scale (medium computation speed), and C = 2 represents a standard scale.

[0136] Based on the duration, scale classification is determined: C=0 represents microscale, with the fastest computation; C=1 represents fine scale, with moderate computation speed; and C=2 represents standard scale. Corresponding multi-scale fusion control is then implemented. Since the fluctuation degree of observation information at different small scales varies, if large fluctuations are unlikely to occur in a short period, the need for long-term prediction is considered relatively small; conversely, the need is considered relatively large. In the case study analysis, only when C=0 or C=1 is a balance between computational power and prediction accuracy established to achieve a balance between prediction and computational power.

[0137] Figure 7 This is a flowchart illustrating a dynamic downscaling-based real-time analysis method according to an embodiment of the present invention, which involves overlaying and displaying the high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data to form a multi-scale nested data display.

[0138] like Figure 7 As shown, in one or more embodiments, preferably, the step of overlaying and displaying the high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data to form multi-scale nested display data specifically includes:

[0139] S701. Extract the high-resolution forecast field, the mesoscale forecast field, and the analysis and forecast data according to a preset period;

[0140] S702. Assimilate the high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data to form an assimilated forecast field;

[0141] S703. Display the data of the assimilation forecast field on a screen.

[0142] In this embodiment of the invention, in order to obtain a close observation and consider the multi-scale nature of the observation, the Spatiotemporal Multiscale Sequential Variational Fusion Analysis (STMAS) algorithm is adopted, which can fuse multiple observations. This analysis method adopts multi-grid technology, and compared with the three-dimensional variational method and the traditional observability analysis method, it has multi-scale characteristics, the analysis increment is anisotropic, and it considers multiple time layers to ensure the temporal continuity of the analysis.

[0143] According to a second aspect of the present invention, a real-time analysis system based on dynamic downscaling is provided.

[0144] Figure 8 This is a structural diagram of a real-world analysis system based on dynamic downscaling, according to an embodiment of the present invention.

[0145] In one or more embodiments, preferably, the real-world analysis system based on dynamic downscaling includes:

[0146] Global model module 801 is used to cold start the global model, extract parameters, and obtain the initial field and boundary field at the current moment as the background field;

[0147] The cold and hot start module 802 is used to hot start the mesoscale mode based on the background field to generate a high-resolution forecast field;

[0148] The mesoscale forecast module 803 is used to perform a 1-kilometer resolution hot start interval to form mesoscale assimilated observation data, and then assimilate it based on the high-resolution forecast field to generate the mesoscale forecast field.

[0149] The small-scale assimilation module 804 is used to perform small-scale analysis according to a preset 1-hour cycle to obtain small-scale parameters.

[0150] The multi-scale fusion module 805 is used to perform multi-scale fusion judgment based on the small-scale parameters to form analysis and forecast data at different scales;

[0151] The rendering and display module 806 is used to overlay and display the high-resolution forecast field, the mesoscale forecast field, and the analysis and forecast data to form multi-scale nested display data.

[0152] In this embodiment of the invention, a system suitable for different structures is realized through a series of modular designs. This system can achieve closed-loop, reliable, and efficient execution through data acquisition, analysis, and control.

[0153] According to a third aspect of the present invention, a computer-readable storage medium is provided that stores computer program instructions thereon, which, when executed by a processor, implement the method as described in any one of the first aspects of the present invention.

[0154] According to a fourth aspect of the present invention, an electronic device is provided. Figure 9 This is a structural diagram of an electronic device according to one embodiment of the present invention. Figure 9 The illustrated electronic device is a general-purpose dynamic downscaling-based real-time analysis device, comprising a general computer hardware architecture, including at least a processor 901 and a memory 902. The processor 901 and memory 902 are connected via a bus 903. The memory 902 is adapted to store instructions or programs executable by the processor 901. The processor 901 can be a standalone microprocessor or a collection of one or more microprocessors. Thus, the processor 901 executes the instructions stored in the memory 902, thereby performing the method flow of the embodiments of the present invention as described above to process data and control other devices. The bus 903 connects the aforementioned components together, and also connects these components to a display controller 904, a display device, and an input / output (I / O) device 905. The input / output (I / O) device 905 can be a mouse, keyboard, modem, network interface, touch input device, motion-sensing input device, printer, and other devices known in the art. Typically, the input / output device 905 is connected to the system via an input / output (I / O) controller 906.

[0155] Figure 10 This outlines the workflow and timeline for nested forecasting and analysis in a downscaling manner. For example... Figure 10 As shown, the specific implementation steps are as follows:

[0156] At 02:00 Beijing Time (18:00 UTC), the global model forecast field serves as the initial field and horizontal lateral boundary field for the mesoscale model. A cold start is performed on a 9 km resolution mesoscale model. Its forecast field serves as the background field for a higher resolution model.

[0157] The 1-kilometer resolution uses a hot-start method, employing 3DVar and cloud analysis methods to assimilate radar and other observational data, absorbing observed small- and medium-scale information, and making short-term forecasts.

[0158] Using a 1-kilometer resolution as the background field, a multi-scale sequential variational method was used to fuse observations from multiple past time periods to obtain the final real-world analysis field.

[0159] Figure 11 This is a schematic diagram of the results based on dynamic downscaling analysis, such as... Figure 11 As shown, this is a meso-β-scale convective weather event that occurred in Liupanshui, Guizhou on September 18, 2021. The strong winds during this event caused casualties. Figure 11The image on the left shows a 10-meter wind field obtained by fusing the global model forecast field as the background field with the data from Chinese ground observation stations. Due to the complex terrain and sparse observation network in this region, the mesoscale weather system in this event could not be distinguished, so the resulting wind field is cyclonic and large in scale. Figure 11 The left side shows the background field, pre-assimilated by radar data using a mesoscale model. The resulting sinking divergence wind field conforms to the circulation characteristics of a convective cell and is located close to the event location. Through numerical simulation downscaling, we can assimilate limited observational data to adjust the thermal and dynamic fields. Then, through model integration, we can force out the small- and mesoscale systems, allowing them to grow and evolve. The resulting simulated forecast field is then fused using a multi-scale fusion method, combining current and past observations to obtain a real-world analysis field containing multi-scale characteristics.

[0160] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0161] This invention's approach uses numerical simulation to compensate for the insufficient spatiotemporal resolution of data from small- and medium-scale weather systems, reconstructing near-surface three-dimensional wind and thermal fields. This is an important tool for studying the dynamics of small- and medium-scale weather systems. Small-scale weather systems like tornadoes, limited by mesoscale atmospheric models and computer capabilities, are primarily studied and used for historical case simulations. Currently, operational short-term forecasting models can achieve a regional 3-kilometer level.

[0162] This invention primarily aims to create real-time analysis fields without requiring long-term forecasts. Therefore, it incorporates small- to medium-scale observations into the model's initial field, enabling model integration to generate small- to medium-scale systems. This allows the model to force out small-scale weather systems with limited observations, providing a background field containing a relatively complete weather system in the real-time analysis. Furthermore, a multi-scale assimilation method is employed to fuse multi-source observation data, particularly radar observations, to improve the three-dimensional thermodynamic structure.

[0163] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0164] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0165] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0166] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0167] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A real-world analysis method based on dynamic downscaling, characterized in that, The method includes: The global model is cold-started to extract parameters and obtain the initial field and boundary field at the current moment, which are used as the background field. Based on the background field, a hot-start mesoscale model is used to generate a high-resolution forecast field; A 1-kilometer resolution hot start interval is used to form mesoscale assimilated observation data, which is then assimilated based on the high-resolution forecast field to generate a mesoscale forecast field. Small-scale analysis is performed according to a preset 1-hour cycle to obtain small-scale parameters; Based on the small-scale parameters, multi-scale fusion judgment is performed to form analysis and forecast data at different scales; The high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data are overlaid and displayed to form a multi-scale nested display data; Specifically, the step of performing multi-scale fusion judgment based on the small-scale parameters to form analysis and forecast data at different scales includes: Based on the small-scale parameters, each parameter is updated once, and the update time is recorded as the single data update time. The scale classification of each parameter is calculated using the first calculation formula; The long-term importance of each parameter is calculated using the second calculation formula; The optimal target prediction count and the optimal target prediction duration are extracted using the third, fourth, and fifth calculation formulas. The calculation is performed when the scale level corresponding to the values ​​of the third and fourth calculation formulas is 1 or 0. When the scale level is 2, the optimal target prediction count for the corresponding parameter is set to a preset fixed value, and the optimal target prediction duration for the corresponding parameter is the product of the second level margin and the preset fixed value. Based on the number of predictions for the optimal target and the prediction duration for the optimal target, analytical forecast data at different scales are generated. The first calculation formula is: in, C To classify the scale. Y 1 represents the first level of margin. Y 2 represents the second level of margin. D This refers to the time between a single data update. The second calculation formula is: in, Z For the aforementioned long-term importance, T The preset fluctuation analysis period, t For time, △ B For parameter fluctuation per unit time, B AVG The mean of the parameters; The third calculation formula is: Y = PJ C in, Y For the prediction accuracy index, J To preset the iteration precision, P Predict the number of times per unit time; The fourth calculation formula is: A = Y / SU p in, A To predict the degree of balance, SU p The number of predictions per unit time is P The corresponding computing power consumption at that time; The fifth calculation formula is: in, A i For the first i The predictive balance of parameters, Z For the first i The long-term importance of parameters, P i For the first i The optimal number of target predictions for the parameters. T Y_i For the first i The optimal target prediction duration for the parameters. SU i For the first i Parameters: computation time per unit of time P i For the first i The number of predictions per unit time parameter S ALL Total computing power per unit time N This represents the total number of parameters.

2. The real-world analysis method based on dynamic downscaling as described in claim 1, characterized in that, The cold-start global model extracts parameters to obtain the initial field and boundary field at the current moment, which serve as the background field. Specifically, this includes: At 2:00 AM Beijing time, global model analysis was initiated, along with initial forecast data. The initial forecast data, with a resolution of 9 kilometers, is used to extract the initial field and the horizontal lateral boundary field as the background field.

3. The field analysis method based on dynamic downscaling as described in claim 1, characterized in that, The step of generating a high-resolution forecast field by hot-starting a mesoscale model based on the background field specifically includes: Based on the aforementioned background field, the hot start mode is activated at 3:00 AM Beijing time; Upon receiving the hot start mode, a hot start mesoscale mode command is issued, and the changes in the current background field are automatically analyzed. The background field at the current moment is used as the forecast field of the previous moment and stored to generate a high-resolution forecast field.

4. The real-world analysis method based on dynamic downscaling as described in claim 1, characterized in that, The process of performing a 1-kilometer resolution warm-start interval to form mesoscale assimilated observation data, and then assimilating it based on the high-resolution forecast field to generate a mesoscale forecast field, specifically includes: Upon receiving the hot-start mesoscale mode command, it performs 3DVar-based data acquisition at a resolution of 1 km to generate radar data. Radar data is directly assimilated to form a synchronous field with complex cloud and water materials, which serves as a mesoscale forecast field.

5. The field analysis method based on dynamic downscaling as described in claim 1, characterized in that, The step of performing small-scale analysis according to a preset 1-hour cycle to obtain small-scale parameters specifically includes: Small-scale analysis is performed according to a preset 1-hour cycle; Obtain all radar data, satellite data, and geographic information to obtain small-scale parameters.

6. The field analysis method based on dynamic downscaling as described in claim 1, characterized in that, The process of overlaying and displaying the high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data to form a multi-scale nested display data specifically includes: According to a preset period, the high-resolution forecast field, the mesoscale forecast field, and the analysis and forecast data are extracted; The high-resolution forecast field, the mesoscale forecast field, and the analyzed forecast data are assimilated to form an assimilated forecast field. The data of the assimilation forecast field is displayed on a single screen.

7. A real-world analysis system based on dynamic downscaling, characterized in that, The system is used to implement the method as described in any one of claims 1-6, the system comprising: The global model module is used to cold start the global model, extract parameters, and obtain the initial field and boundary field at the current moment as the background field. A cold-hot start module is used to hot-start a mesoscale model based on the background field to generate a high-resolution forecast field; The mesoscale forecast module is used to perform a 1-kilometer resolution hot start interval to form mesoscale assimilated observation data, and then assimilate it based on the high-resolution forecast field to generate the mesoscale forecast field. The small-scale assimilation module is used to perform small-scale analysis according to a preset 1-hour cycle to obtain small-scale parameters. The multi-scale fusion module is used to perform multi-scale fusion judgment based on the small-scale parameters to form analysis and forecast data at different scales; The rendering and display module is used to overlay and display the high-resolution forecast field, the mesoscale forecast field, and the analysis and forecast data to form multi-scale nested display data.

8. A computer-readable storage medium storing computer program instructions thereon, characterized in that, The computer program instructions, when executed by a processor, implement the method as described in any one of claims 1-6.

9. An electronic device comprising a memory and a processor, characterized in that, The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described in any one of claims 1-6.

Images