Unmanned aerial vehicle under-throw sounding data assimilation method and device, electronic equipment and medium

Abstract

Embodiments of the present application relate to the technical field of meteorological observation and data assimilation, and provide a UAV dropsonde data assimilation method and device, an electronic device and a medium. Raw profile data collected when a UAV dropsonde is performed is acquired. The overall variance of the raw profile data is calculated along the direction of the atmospheric vertical section. The raw profile data is vertically sampled based on the overall variance and a preset variance ratio criterion to obtain target sampled profile data. The variance ratio of the target sampled profile data relative to the overall variance is maximum, and the number of sampling layers matches the number of vertical layers of a global forecast system data background field. The target sampled profile data and the global forecast system data background field are assimilated to obtain an assimilated model meteorological field. Thus, the observation data is down-sampled, the original data information is ensured, the observation noise is removed, and the assimilation effect and the prediction accuracy are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of meteorological observation and data assimilation, and particularly relates to a UAV dropsonde data assimilation method and device, an electronic equipment and a medium. BACKGROUND

[0002] Meteorological observation is the basis of weather forecasting, not only providing support for accurate forecasting, but also being a key way to promote the continuous development of atmospheric science research and weather forecasting technology. In recent years, with the increase of observation sites and the development of assimilation technology, the quality of numerical prediction has been significantly improved. However, there is still a large error in the prediction of complex terrain by the current numerical prediction model, especially in the plateau area where the observation sites are sparse. For improving the prediction results in complex terrain areas, observation data is essential. Because the simulation results of the numerical model are highly dependent on the initial state, and the accuracy of the initial state is directly determined by the meteorological observation data.

[0003] For the plateau area, the population is sparse, and the transportation, communication and other infrastructure are weak, so the observation sites are few, and the effective sounding observation data obtained only includes balloon sounding and satellite remote sensing. Although balloon sounding can directly measure meteorological elements at different altitudes and fully present the vertical structure of the atmosphere, its horizontal representativeness is limited, and it may be interrupted due to weather. Compared with balloon sounding, satellite remote sensing can quickly obtain large-area observation data and break through the spatial limitations of ground observation, but there is a large error in the data inversion. Based on this, in recent years, a kind of downcast sounding using an airplane has been widely used, which has high mobility. It can launch a downcast sonde in the target observation area, at a specific time, at a specified height and with the required spatial density, so as to obtain high-precision, high-spatial-resolution three-dimensional meteorological observation data. Assimilating these detection results into the numerical prediction model can significantly improve the accuracy of meteorological prediction. The UAV drop test usually launches a downcast sonde in the middle troposphere (about 400 hPa) to carry out sounding work, so as to detect the wind field, temperature field and humidity field profile of the atmosphere below 400 hPa. A large number of studies have shown that it is important for the prediction of tropical cyclone path and intensity, and it is an internationally recognized advanced observation method for improving the simulation accuracy of cyclones. However, there is still a lack of consensus on how to effectively introduce UAV mobile detection data into the numerical assimilation system in complex terrain environment, and then improve the prediction simulation in mountainous areas.

[0004] Downward projection observations offer high vertical resolution, with up to nearly a thousand layers observed at different altitudes from the start to the end of the projection process. This means a more comprehensive reflection of the three-dimensional structure of the atmosphere. During assimilation, the more layers of observational data incorporated, the more detailed the description of the true vertical structure of the atmosphere, allowing for more accurate adjustment of the model's initial field. Studies have shown that observational data from wind profiler radars and ground-based microwave radiometers at different levels significantly impact the forecasting performance of numerical weather prediction (WRF) models. However, more layers are not always better during assimilation. If the quality of the assimilated observational data is low, more layers may introduce more "noise." Furthermore, current numerical weather prediction models only have a few dozen layers vertically. Increasing to hundreds of layers with observational data would significantly increase computational costs and potentially lead to numerical instability or background field imbalance. How to assimilate more observational data while maintaining the original model layers is a pressing issue that needs to be addressed.

[0005] Currently, due to the efficiency requirements of daily numerical weather prediction operations, the number of vertical layers is far less than that of dropsonde observations. This leads to a mismatch in vertical resolution, and direct assimilation without processing results in multiple observation points corresponding to the same model layer, causing information redundancy. Furthermore, a large amount of high-frequency vertical information cannot be resolved by the model, easily causing numerical instability, background field imbalance, and even assimilation divergence. In addition, excessively dense observation data increases the computational burden on the assimilation system, reducing efficiency. Therefore, how to downsample the observation data for effective information assimilation has become an urgent problem to be solved. Summary of the Invention

[0006] This invention provides a method, apparatus, electronic device, and medium for assimilating UAV-dropped radiosonde data. It addresses the shortcomings of existing technologies where numerical weather prediction has far fewer vertical layers than drop radiosonde observations, resulting in a mismatch in vertical resolution and redundant drop radiosonde information that increases the computational burden on the assimilation system. The invention achieves downsampling of observation data, preserving the original data information while removing observation noise, thereby improving the assimilation effect and forecast accuracy.

[0007] This invention provides a method for assimilating UAV-dropped radiosonde data, comprising:

[0008] Acquire raw profile data collected during UAV-dropped sounding;

[0009] Calculate the overall variance of the original profile data along the direction of the vertical atmospheric profile;

[0010] sampling the original profile data based on the total variance and a preset variance ratio criterion to obtain target sampling profile data, a variance ratio of the target sampling profile data relative to the total variance being maximum and a number of sampling layers matching a number of vertical layers of a global forecast system data background field;

[0011] assimilating the target sampling profile data with the global forecast system data background field to obtain an assimilated model meteorological field.

[0012] In one possible implementation, the method further includes:

[0013] selecting different sampling layer combinations, and calculating a variance ratio of a variance of each sampling layer combination to the total variance;

[0014] taking a sampling layer combination that satisfies the preset variance ratio criterion and has the least number of sampling layers as a target sampling layer combination, wherein the preset variance ratio criterion is that the variance ratio is greater than or equal to a preset threshold value;

[0015] sampling the original profile data based on the target sampling layer combination to obtain target sampling profile data.

[0016] In one possible implementation, the method further includes:

[0017] inputting the sampling profile data and the global forecast system data background field into an assimilation system, and assimilating, by the assimilation system, the target sampling profile data with the global forecast system data background field based on an optimal interpolation algorithm to obtain an assimilated model meteorological field.

[0018] In one possible implementation, the method further includes:

[0019] inputting the assimilated model meteorological field into a numerical prediction model to drive the numerical prediction model to perform numerical weather prediction for a future period.

[0020] In one possible implementation, the method further includes:

[0021] connecting the assimilation system with the numerical prediction model through a standardized interface;

[0022] inputting the assimilated model meteorological field into the numerical prediction model based on the standardized interface to match a resolution of the assimilated model meteorological field with an initial field of the numerical prediction model;

[0023] performing numerical weather prediction for a future period by the numerical prediction model with the matched resolution.

[0024] In one possible implementation, the method further includes:

[0025] The numerical weather forecast is evaluated based on the weather forecast data obtained from the model meteorological fields corresponding to the assimilated first and second sampling frequencies, respectively. The evaluation assesses the improvement effect of the assimilated model meteorological fields on multiple data parameters of the numerical weather forecast, including at least two of precipitation location and intensity, near-surface air temperature, boundary layer structure, and computational stability.

[0026] The present invention also provides a device for assimilating radiosonde data dropped by an unmanned aerial vehicle (UAV), comprising the following modules:

[0027] The acquisition module is used to acquire raw profile data collected during UAV-dropped sounding.

[0028] The calculation module is used to calculate the overall variance of the original profile data along the direction of the atmospheric vertical profile.

[0029] The sampling module is used to perform vertical sampling on the original profile data based on the total variance and a preset variance ratio criterion to obtain target sampled profile data. The variance ratio of the target sampled profile data to the total variance is the largest, and the number of sampling layers matches the number of vertical layers of the background field of the global forecast system data.

[0030] The assimilation module is used to assimilate the target sampling profile data with the background field of the global forecast system data to obtain the assimilated model meteorological field.

[0031] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the UAV-dropped radiosonde data assimilation method as described above.

[0032] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the UAV-dropped sounding data assimilation method as described above.

[0033] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the UAV-dropped radiosonde data assimilation method as described above.

[0034] This invention provides a method, apparatus, electronic device, and medium for assimilating UAV-dropped radiosonde data. The method involves acquiring raw profile data collected during UAV-dropped radiosonde observations; calculating the overall variance of the raw profile data along the vertical atmospheric profile; vertically sampling the raw profile data based on the overall variance and a preset variance ratio criterion to obtain target sampled profile data. The target sampled profile data has the largest variance ratio relative to the overall variance, and its sampling layer number matches the vertical layer number of the global forecast system's background data field. The target sampled profile data is then assimilated with the global forecast system's background data field to obtain an assimilated model meteorological field. Compared to existing technologies where numerical weather prediction has far fewer vertical layers than drop-sonde observations, resulting in a mismatch in vertical resolution and redundant drop-sonde information that increases the computational burden on the assimilation system, this solution achieves downsampling of the observation data, preserving the original data information while removing observation noise, thus improving the assimilation effect and forecast accuracy. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0036] Figure 1 This is one of the flowcharts of the UAV-dropped sounding data assimilation method provided by the present invention.

[0037] Figure 2 This is the second flowchart of the UAV-dropped sounding data assimilation method provided by the present invention.

[0038] Figure 3 This is the third flowchart of the UAV-dropped sounding data assimilation method provided by the present invention.

[0039] Figure 4 This is a data sampling flowchart provided by the present invention.

[0040] Figure 5 This is a schematic diagram of the variance ratio under different vertical sampling intervals provided by the present invention.

[0041] Figure 6 This is a schematic diagram of the average root mean square error of the hourly 2-meter temperature forecast for the Sichuan region provided by this invention.

[0042] Figure 7 This is a schematic diagram of the ETS score for the 24-hour cumulative precipitation forecast of Chengdu and its surrounding areas provided by the present invention.

[0043] Figure 8 This is a schematic diagram of the structure of the UAV-dropped radiosonde data assimilation device provided by the present invention.

[0044] Figure 9 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0046] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0047] Figure 1 This is one of the flowcharts illustrating the UAV-based assimilation method for sounding data from aerial radiosondes provided by this invention. Figure 1 As shown, the method includes the following:

[0048] S11. Obtain the raw profile data collected during the UAV's airborne sounding.

[0049] In this embodiment of the invention, a drop radiosonde test is performed in the target area by an unmanned aerial vehicle (UAV) to obtain drop radiosonde observation data transmitted back in real time after the radiosonde is deployed. The drop radiosonde observation data includes raw profile data.

[0050] A single profile typically includes elements such as air pressure, temperature, humidity, wind speed, and wind direction. For example, the sampling interval along the height direction can reach 5–10 m, with a total of nearly 1000 layers.

[0051] Optionally, the raw profile data can be preprocessed to remove obvious outliers (such as temperature-dew point difference >30℃, wind speed >100 m / s, etc., which are physically unreasonable values). Unify the altitude coordinates: convert GPS altitude to geopotential or barometric altitude to facilitate subsequent alignment with the background field of the Global Forecast System (GFS). This data preprocessing provides clean, isotropic raw data for subsequent calculations of the overall variance and variance ratio sampling.

[0052] S12. Calculate the overall variance of the original profile data along the direction of the vertical atmospheric profile.

[0053] Based on the original profile data, the overall variance of variables such as air pressure, temperature, humidity, wind speed, and wind direction in the vertical profile is calculated, serving as a quantitative benchmark for the complexity and information content of the original vertical structure.

[0054] S13. Based on the total variance and the preset variance ratio criterion, the original profile data is vertically sampled to obtain the target sampled profile data.

[0055] The goal of this step is to find a vertical sampling scheme in which the variance of the selected sampling profile data is maximized relative to the overall variance, and the number of sampling layers matches the number of vertical layers in the background field of the global forecast system data.

[0056] Specifically, different sampling layer combinations are selected, and the variance ratio of each sampling layer combination to the overall variance is calculated. The sampling layer combination that satisfies the preset variance ratio criterion and has the fewest sampling layers is selected as the target sampling layer combination, wherein the preset variance ratio criterion is that the variance ratio is greater than or equal to a preset threshold. Based on the target sampling layer combination, sampling is performed on the original profile data to obtain the target sampling profile data.

[0057] S14. Assimilate the target sampling profile data with the background field of the global forecast system data to obtain the assimilated model meteorological field.

[0058] The target sampled profile data and the background field of the Global Forecast System (GFS) data are input into the assimilation system. The assimilation system uses the optimal interpolation algorithm to assimilate the target sampled profile data and the background field of the GFS data to obtain the assimilated model meteorological field.

[0059] Furthermore, the assimilated model meteorological field can be input into the numerical weather prediction model to drive the numerical weather prediction model to make numerical weather predictions for future periods.

[0060] The UAV-based dropsonde data assimilation method provided by this invention acquires raw profile data collected during UAV dropsonde observations; calculates the overall variance of the raw profile data along the vertical atmospheric profile; performs vertical sampling on the raw profile data based on the overall variance and a preset variance ratio criterion to obtain target sampled profile data. The target sampled profile data has the largest variance ratio relative to the overall variance, and its sampling layer number matches the vertical layer number of the global forecast system background field. The target sampled profile data is then assimilated with the global forecast system background field to obtain the assimilated model meteorological field. Compared to existing technologies where numerical weather prediction has far fewer vertical layers than dropsonde observations, resulting in a mismatch in vertical resolution and redundant dropsonde information that increases the computational burden on the assimilation system, this method achieves downsampling of the observation data, preserving the original data information while removing observation noise, thus improving the assimilation effect and forecast accuracy.

[0061] Figure 2 This is the second flowchart illustrating the UAV-based assimilation method for sounding data from aerial radiosondes provided by this invention. Figure 2 As shown, the method includes the following:

[0062] S21. Obtain the raw profile data collected by the UAV during the airborne sounding, and calculate the overall variance of the raw profile data along the direction of the vertical atmospheric profile.

[0063] The embodiments of the present invention are combined with Figure 3 The flowchart illustrating the UAV-dropped radiosonde data assimilation method is explained in detail.

[0064] The drone flies within the operational area (e.g., a 500 km x 500 km area in the western Sichuan plateau), and upon reaching the pre-set drop point, releases a drop-sonde in a single operation. The radiosonde transmits data every 5 seconds, from 400 hPa down to the ground, forming a "super-high resolution" raw profile—nearly a thousand recording points for temperature, humidity, wind speed, wind direction, and air pressure as altitude changes. These recording points are treated as a whole, and the overall variance of each variable is calculated. A larger value indicates a more tortuous vertical structure and richer meteorological information within the profile; a smaller value indicates a smoother profile. This overall variance serves as the benchmark for all subsequent sampling schemes.

[0065] S22. Select different sampling layer combinations and calculate the variance ratio of each sampling layer combination to the total variance.

[0066] In this embodiment of the invention, a variety of possible sampling intervals can be automatically generated using computer scripts: 50 m, 100 m, 150 m... 1000 m, with dozens of combinations. For each selected interval, only the layers corresponding to the height are retained in the original profile, and all other data are discarded; then, the variance is recalculated at the retained points. The "retained variance" is divided by the "overall variance" to obtain the "variance ratio." This results in an "interval-variance ratio" curve, as shown in Figure 5, where the horizontal axis represents the number of sampling layers (or intervals), and the vertical axis represents the variance ratio.

[0067] S23. The sampling layer combination that satisfies the preset variance ratio criterion and has the fewest sampling layers is taken as the target sampling layer combination.

[0068] The preset variance ratio criterion is that the variance ratio is greater than or equal to a preset threshold.

[0069] Set an empirical threshold: variance ratio ≥ 0.9 (or 0.95, which can be adjusted according to business tolerance). On the graph, find the first point from left to right that meets the threshold and has the fewest sampling layers; this is the "optimal compromise point". Experiments show that a 250 m interval perfectly satisfies "variance ratio 0.93, number of layers ≈ 40", so 250 m is chosen as the final sampling interval, and the corresponding height set is the "target sampling layer combination".

[0070] S24. Based on the target sampling layer combination, sample the original profile data to obtain target sampling profile data.

[0071] Returning to the original profile, only the height points in the target layer combination are extracted to form a "simplified" profile. The simplified profile data volume is compressed from nearly a thousand layers to about 40 layers, but more than 90% of the key fluctuation information is retained. At the same time, the number of layers basically matches the number of vertical layers of the GFS background field, which can be directly used in the assimilation system to avoid information redundancy and numerical instability.

[0072] S25. Input the sampled profile data and the background field of the global forecast system into the assimilation system. The assimilation system uses the optimal interpolation algorithm to assimilate the target sampled profile data and the background field of the global forecast system to obtain the assimilated model meteorological field.

[0073] Using the GFS global forecast field as the background field and the 40-layer sampling profile obtained in the previous step as the observation input, both are fed into the ADAS assimilation system.

[0074] The ADAS assimilation system uses an optimal interpolation algorithm: first, it calculates the difference between the observation and the background field, and then assimilates the difference to the background field according to statistical weights to obtain the "assimilated model meteorological field". After this step, the temperature, humidity, and wind field in the background field have been corrected by the actual observation of the UAV, especially in key areas such as the low-level jet stream, inversion layer, and humid layer.

[0075] S26. Input the assimilated model meteorological field into the numerical weather prediction model to drive the numerical weather prediction model to perform numerical weather forecasts for future periods.

[0076] The assimilated model meteorological field is used as the initial condition for the WRF model, and a 24-hour integration is initiated. The WRF model outputs hourly 2-meter air temperature, 10-meter wind, and 3-km resolution precipitation, among other conventional meteorological elements, forming the final numerical weather prediction product.

[0077] S27. The numerical weather forecast is evaluated based on the weather forecast data obtained from the model meteorological fields corresponding to the assimilated first and second sampling frequencies, respectively, to evaluate the improvement effect of the assimilated model meteorological fields on multiple data parameters of the numerical weather forecast.

[0078] Multiple data parameters include at least two of precipitation location and intensity, near-surface air temperature, boundary layer structure, and computational stability.

[0079] After employing "variance ratio control + ADAS assimilation", UAV-based dropsonde significantly improved the accuracy of short-term weather forecasts in the complex terrain of Southwest China, while the increase in computational load remained manageable and could be stably operated in the operational system.

[0080] Specifically, to evaluate the reliability of the above method, the UAV-dropped radiosonde data was sampled at equal altitude intervals, and then... Figure 4 The method shown demonstrates that when the height sampling interval is set to 250m (the first sampling frequency), the original information variance ratio can be maximized while ensuring that the number of layers in the original dropsonde data is not too dense. Figure 5 Based on this, a 24-hour forecast experiment was conducted. The experiment used WRF (Weather Research and Forecasting Model) as the numerical weather prediction model. To demonstrate the effectiveness of this method, two sets of experiments were run: one was the DROP_250m experiment with an assimilation vertical altitude interval of 250m, and the other was the DROP_650m experiment with a vertical altitude interval of 650m (second sampling frequency). The forecast results were then compared.

[0081] like Figure 6As shown, the root mean square (RMSE) of hourly 2-meter temperature forecasts in Sichuan Province was analyzed for both experimental groups. The results show that the DROP_250m experiment with a vertical assimilation interval of 250m exhibited lower errors in most time intervals, especially from afternoon to night, with a significantly lower RMSE than the 650m scheme. This indicates that higher vertical resolution helps the model more accurately characterize surface temperature changes, thereby improving the accuracy of near-surface temperature forecasts. Overall, the refined vertical assimilation interval significantly enhances the model's response to boundary layer thermal structure and demonstrates superior forecasting skill.

[0082] Figure 7 This paper presents a comparative analysis of the Equitable Threat Score (ETS) of two sets of experiments in precipitation forecasting. Compared to the DROP_650m experiment with a vertical assimilation interval of 650m, the DROP_250m experiment (250m), with a denser vertical interval, showed better overall scores at precipitation thresholds of 10.0mm, 25.0mm, 50.0mm, and 60.0mm. This is because the 250m interval scheme can more finely characterize the vertical structure of the atmosphere (such as the distribution of the moist layer, upward motion, and convection development processes), significantly improving its ETS score, especially showing a more obvious advantage in forecasting moderate to heavy rainfall (particularly heavy precipitation). This result indicates that enhanced vertical resolution helps improve the precipitation forecasting skill of numerical models.

[0083] This invention innovatively proposes a sampling method based on the "variance ratio" criterion. It selects the most representative layers with optimal information retention from nearly a thousand layers of original airborne sounding data, preserving the main variation features of the original profile with as few layers as possible, ensuring that key meteorological structures (such as fronts, inversion layers, and wet layers) are not weakened or lost. Addressing the vertical inconsistency between sounding data and numerical weather prediction models, a complete and efficient processing workflow from sampling and background field construction to data assimilation is designed. This achieves seamless integration of high-density observational information with low-resolution model structures, ensuring information utilization while avoiding numerical instability. By integrating the processed observational data into the ADAS assimilation system and applying it to the WRF forecasting model, experimental results show significant improvements in forecast accuracy for boundary layer structure, precipitation location, and magnitude, demonstrating good operational promotion value and practical application potential. After integrating this method into the ADAS assimilation system and applying it to the WRF weather forecast model, the experimental results show that the simulation of precipitation location and intensity is more accurate, especially under sensitive weather backgrounds such as short-duration heavy precipitation and plateau boundary layer disturbance, the forecast is significantly improved.

[0084] The UAV-dropped radiosonde data assimilation device provided by the present invention is described below. The UAV-dropped radiosonde data assimilation device described below and the UAV-dropped radiosonde data assimilation method described above can be referred to in correspondence with each other.

[0085] Figure 8 This is a schematic diagram of the structure of the UAV-dropped radiosonde data assimilation device provided by the present invention, specifically including:

[0086] The acquisition module 801 is used to acquire the raw profile data collected when the UAV drops a sounding point.

[0087] Calculation module 802 is used to calculate the overall variance of the original profile data along the direction of the atmospheric vertical profile;

[0088] The sampling module 803 is used to perform vertical sampling on the original profile data based on the total variance and a preset variance ratio criterion to obtain target sampled profile data. The variance ratio of the target sampled profile data to the total variance is the largest, and the number of sampling layers matches the number of vertical layers of the background field of the global forecast system data.

[0089] The assimilation module 804 is used to assimilate the target sampling profile data with the background field of the global forecast system data to obtain the assimilated model meteorological field.

[0090] In one possible implementation, the sampling module 803 is further configured to select different sampling layer combinations, calculate the variance ratio of each sampling layer combination to the overall variance, and select the sampling layer combination that satisfies the preset variance ratio criterion and has the fewest sampling layers as the target sampling layer combination, wherein the preset variance ratio criterion is that the variance ratio is greater than or equal to a preset threshold; and sample the original profile data based on the target sampling layer combination to obtain the target sampling profile data.

[0091] In one possible implementation, the assimilation module 804 is further configured to input the sampled profile data and the background field of the global forecast system into the assimilation system, and then use the assimilation system to assimilate the target sampled profile data and the background field of the global forecast system based on the optimal interpolation algorithm to obtain the assimilated model meteorological field.

[0092] In one possible implementation, the assimilation module 804 is further configured to input the assimilated model meteorological field into the numerical weather prediction model, thereby driving the numerical weather prediction model to perform numerical weather forecasts for future periods.

[0093] In one possible implementation, the assimilation module 804 is further configured to interface the assimilation system with a numerical weather prediction model through a standardized interface; input the assimilated model meteorological field into the numerical weather prediction model based on the standardized interface; match the resolution of the assimilated model meteorological field with the initial field of the numerical weather prediction model; and perform numerical weather forecasts for future periods using the numerical weather prediction model with the matched resolution.

[0094] In one possible implementation, the calculation module 802 is further configured to evaluate the numerical weather forecast based on weather forecast data obtained from model meteorological fields corresponding to the assimilated first and second sampling frequencies, respectively, and to evaluate the improvement effect of the assimilated model meteorological fields on multiple data parameters of the numerical weather forecast, wherein the multiple data parameters include at least two of precipitation location and intensity, near-surface air temperature, boundary layer structure, and computational stability.

[0095] Figure 9 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 9 As shown, the electronic device may include: a processor 910, a communication interface 920, a memory 930, and a communication bus 940, wherein the processor 910, the communication interface 920, and the memory 930 communicate with each other through the communication bus 940. The processor 910 can call logical instructions in the memory 930 to execute a method for assimilating UAV-dropped radiosonde data. This method includes: acquiring raw profile data collected during UAV-dropped radiosonde; calculating the overall variance of the raw profile data along the direction of the atmospheric vertical profile; vertically sampling the raw profile data based on the overall variance and a preset variance ratio criterion to obtain target sampled profile data, wherein the variance of the target sampled profile data has the largest variance ratio relative to the overall variance and the number of sampling layers matches the number of vertical layers of the global forecast system background field; and assimilating the target sampled profile data with the global forecast system background field to obtain an assimilated model meteorological field.

[0096] Furthermore, the logical instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the UAV dropsonde data assimilation method provided by the above methods. The method includes: acquiring raw profile data collected during UAV dropsonde; calculating the overall variance of the raw profile data along the direction of the atmospheric vertical profile; vertically sampling the raw profile data based on the overall variance and a preset variance ratio criterion to obtain target sampled profile data, wherein the variance of the target sampled profile data has the largest variance ratio relative to the overall variance and the number of sampling layers matches the number of vertical layers of the global forecast system background field; and assimilating the target sampled profile data with the global forecast system background field to obtain an assimilated model meteorological field.

[0098] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the UAV-dropped radiosonde data assimilation method provided by the above methods. The method includes: acquiring raw profile data collected during UAV-dropped radiosonde; calculating the overall variance of the raw profile data along the direction of the atmospheric vertical profile; vertically sampling the raw profile data based on the overall variance and a preset variance ratio criterion to obtain target sampled profile data, wherein the variance of the target sampled profile data has the largest variance ratio relative to the overall variance and the number of sampling layers matches the number of vertical layers of the global forecast system background field; and assimilating the target sampled profile data with the global forecast system background field to obtain an assimilated model meteorological field.

[0099] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0100] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for assimilating UAV-dropped radiosonde data, characterized in that, include: Acquire raw profile data collected during UAV-dropped sounding; Calculate the overall variance of the original profile data along the direction of the vertical atmospheric profile; Based on the total variance and the preset variance ratio criterion, the original profile data is vertically sampled to obtain target sampled profile data. The variance ratio of the target sampled profile data to the total variance is the largest, and the number of sampling layers matches the number of vertical layers of the background field of the global forecast system data. The process of vertically sampling the original profile data based on the overall variance and a preset variance ratio criterion to obtain sampled profile data includes: Select different combinations of sampling layers and calculate the variance ratio of each sampling layer combination to the total variance. The sampling layer combination that satisfies the preset variance ratio criterion and has the fewest sampling layers is taken as the target sampling layer combination, wherein the preset variance ratio criterion is that the variance ratio is greater than or equal to a preset threshold. Based on the target sampling layer combination, sampling is performed on the original profile data to obtain target sampling profile data; The target sampling profile data is assimilated with the background field of the global forecast system data to obtain the assimilated model meteorological field. The process of assimilating the target sampling profile data with the background field of the global forecast system data to obtain the assimilated model meteorological field includes: The sampled profile data and the background field of the global forecast system are input into the assimilation system. The assimilation system uses the optimal interpolation algorithm to assimilate the target sampled profile data and the background field of the global forecast system to obtain the assimilated model meteorological field.

2. The method according to claim 1, characterized in that, The method further includes: The assimilated model meteorological field is input into the numerical weather prediction model to drive the numerical weather prediction model to make numerical weather predictions for future periods.

3. The method according to claim 2, characterized in that, The step of inputting the assimilated model meteorological field into the numerical weather prediction model to drive the numerical weather prediction model to perform numerical weather prediction for future periods includes: The assimilation system will be integrated with numerical weather prediction models through a standardized interface; Based on the standardized interface, the assimilated model meteorological field is input into the numerical weather prediction model to match the resolution of the assimilated model meteorological field with the initial field of the numerical weather prediction model. Numerical weather forecasts for future periods are made using numerical forecast models with matched resolution.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: The numerical weather forecast is evaluated based on the weather forecast data obtained from the model meteorological fields corresponding to the assimilated first and second sampling frequencies, respectively. The evaluation assesses the improvement effect of the assimilated model meteorological fields on multiple data parameters of the numerical weather forecast, including at least two of precipitation location and intensity, near-surface air temperature, boundary layer structure, and computational stability.

5. A device for assimilating radiosonde data dropped by an unmanned aerial vehicle (UAV), characterized in that, include: The acquisition module is used to acquire raw profile data collected during UAV-dropped sounding. The calculation module is used to calculate the overall variance of the original profile data along the direction of the atmospheric vertical profile. The sampling module is used to perform vertical sampling on the original profile data based on the total variance and a preset variance ratio criterion to obtain target sampled profile data. The variance ratio of the target sampled profile data to the total variance is the largest, and the number of sampling layers matches the number of vertical layers of the background field of the global forecast system data. The step of vertically sampling the original profile data based on the overall variance and a preset variance ratio criterion to obtain sampled profile data includes: selecting different sampling layer combinations and calculating the variance ratio of each sampling layer combination to the overall variance; selecting the sampling layer combination that satisfies the preset variance ratio criterion and has the fewest sampling layers as the target sampling layer combination, wherein the preset variance ratio criterion is that the variance ratio is greater than or equal to a preset threshold; and sampling the original profile data based on the target sampling layer combination to obtain target sampled profile data. An assimilation module is used to assimilate the target sampled profile data with the background field of the global forecast system data to obtain an assimilated model meteorological field. The assimilation of the target sampled profile data with the background field of the global forecast system data to obtain the assimilated model meteorological field includes: inputting the sampled profile data and the background field of the global forecast system data into the assimilation system, and using the assimilation system to assimilate the target sampled profile data with the background field of the global forecast system data based on an optimal interpolation algorithm to obtain the assimilated model meteorological field.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the UAV-dropped sounding data assimilation method as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the UAV-dropped sounding data assimilation method as described in any one of claims 1 to 4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the UAV-dropped sounding data assimilation method as described in any one of claims 1 to 4.

Images