Deep learning-based convection cloud rain reduction operation analysis method and system

Abstract

The invention provides a convective cloud rain reduction operation analysis method and system based on deep learning, and is used for obtaining a comprehensive convective cloud radar echo extrapolation data set, carrying out the innovative data processing, precisely generating an extrapolation result, formulating a reasonable operation strategy, and improving the working efficiency. The problems of insufficient data utilization, unscientific operation strategy formulation and the like in the flow cloud rain reduction operation in the prior art are effectively solved, and the accuracy and effectiveness of the operation are remarkably improved. The method comprises the following steps: carrying out sparse sampling and space-time alignment processing on a convective cloud radar echo extrapolation data set of a target area to generate a space-time continuous radar echo extrapolation reconstruction data set; calling a space-time sequence prediction model to perform cyclic radar echo extrapolation processing on the radar echo extrapolation reconstruction data set, and generating an extrapolation result within a preset time range; and generating a convective cloud rain reduction operation strategy based on the extrapolation result, and sending the convective cloud rain reduction operation strategy to a rain reduction operation service system to execute operation processing.

Description

Technical Field

[0001] This application belongs to the technical field of meteorological processing, and particularly relates to a convective cloud rain attenuation operation analysis method and system based on deep learning. Background Art

[0002] In the field of meteorology, convective cloud rain attenuation operations are of great significance for ensuring the smooth progress of major events and reducing natural disaster losses. With the development of meteorological science and technology, people's requirements for the accuracy and effectiveness of rain attenuation operations are increasing day by day. Traditional convective cloud rain attenuation operations often lack comprehensive and accurate data analysis of convective clouds. Specifically, previous operation decisions mostly relied on experience and simple observation data, and failed to make full use of the complex and changeable convective cloud information; in terms of data processing, it was difficult to effectively integrate the information of multiple time steps for convective cloud radar echo data, and the data coherence was poor, resulting in inaccurate judgments on the development trend of convective clouds. In terms of operation strategy formulation, it was impossible to accurately determine the operation point location and operation timing, and there was a lack of scientific consideration of the operation status evaluation characteristics. Summary of the Invention

[0003] This application provides a convective cloud rain attenuation operation analysis method and system based on deep learning, which can obtain a comprehensive convective cloud radar echo extrapolation data set, perform innovative data processing, accurately generate extrapolation results and formulate reasonable operation strategies, effectively solve the problems of insufficient data utilization and unscientific operation strategy formulation in the existing technology for convective cloud rain attenuation operations, and significantly improve the accuracy and effectiveness of operations.

[0004] In a first aspect, an embodiment of this application provides a convective cloud rain attenuation operation analysis method based on deep learning, which is applied to a convective cloud rain attenuation operation analysis system. The method includes: obtaining a convective cloud radar echo extrapolation data set of a target area, where the convective cloud radar echo extrapolation data set includes radar echo intensity distribution data of multiple time steps and corresponding convective cloud spatial distribution data; performing sparse sampling and spatio-temporal alignment processing on the convective cloud radar echo extrapolation data set to generate a spatio-temporally continuous radar echo extrapolation reconstruction data set; calling a spatio-temporal sequence prediction model to perform cyclic radar echo extrapolation processing on the radar echo extrapolation reconstruction data set to generate extrapolation results within a preset time range, where the extrapolation results include the spatial evolution path and intensity change trend of convective clouds; generating a convective cloud rain attenuation operation strategy based on the extrapolation results, and sending the convective cloud rain attenuation operation strategy to a rain attenuation operation service system for operation processing. The convective cloud rain attenuation operation strategy includes operation point location, operation timing, and operation status evaluation characteristics.

[0005] In a second aspect, an analysis system for convective cloud rain attenuation operations provided by an embodiment of the present application includes a processor and a memory. Among them, the memory stores a computer program. When the computer program is executed by the processor, the processor is caused to execute the steps of the above method.

[0006] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, which includes a computer program. When the computer program runs on an analysis system for convective cloud rain attenuation operations, the computer program is used to cause the analysis system for convective cloud rain attenuation operations to execute the steps of the above method.

[0007] In the implementation of the present application, on the one hand, by performing sparse sampling and spatio-temporal alignment processing on the convective cloud radar echo extrapolation data set of the target area, data can be effectively integrated to generate a spatio-temporally continuous radar echo extrapolation reconstruction data set, greatly improving the coherence and availability of the data, and enabling the radar echo extrapolation reconstruction data set to more accurately reflect the actual situation of convective clouds; on the other hand, by calling a spatio-temporal sequence prediction model for cyclic radar echo extrapolation processing, the potential laws in the data can be deeply mined, and extrapolation results including the spatial evolution path and intensity change trend of convective clouds within a preset time range can be accurately generated, providing a strong basis for subsequent decision-making; on the third hand, the convective cloud rain attenuation operation strategy generated based on the extrapolation results can comprehensively consider the location of the operation point, the operation timing, and the operation status evaluation characteristics, making the rain attenuation operation more targeted and scientific. Sending the convective cloud rain attenuation operation strategy to the rain attenuation operation service system for operation processing can effectively optimize the operation process, improve the accuracy and efficiency of the rain attenuation operation, and provide reliable support for meteorological operations.

[0008] In summary, the embodiment of the present application obtains a comprehensive convective cloud radar echo extrapolation data set, performs innovative data processing, accurately generates extrapolation results and formulates reasonable operation strategies, effectively solving the problems of insufficient data utilization and unscientific operation strategy formulation in the prior art for convective cloud rain attenuation operations, and significantly improving the accuracy and effectiveness of the operations. Description of the Drawings

[0009] Figure 1 It is a schematic flow chart of a method for analyzing convective cloud rain attenuation operations based on deep learning provided by an embodiment of the present application.

[0010] Figure 2 It is a schematic structural diagram of an analysis system for convective cloud rain attenuation operations provided by an embodiment of the present application. Detailed Embodiments

[0011] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions of this application in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are some, rather than all, of the embodiments of the technical solutions of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments recorded in this application document without creative efforts belong to the scope protected by the technical solutions of this application.

[0012] Refer to Figure 1 , which is a method for analyzing convective cloud rain attenuation operations based on deep learning provided in the embodiments of this application. This method can be applied to a convective cloud rain attenuation operation analysis system, and the specific process is as shown in steps 110 - 140.

[0013] Step 110: Obtain an extrapolation dataset of convective cloud radar echoes in the target area. The extrapolation dataset of convective cloud radar echoes includes radar echo intensity distribution data and corresponding convective cloud spatial distribution data at multiple time steps.

[0014] In the embodiments of this application, taking a meteorological monitoring area as an example, multiple radar monitoring stations are set up in this area, and these radar monitoring stations continuously monitor convective clouds. Within a period of time, the radar monitoring stations collect data at set time intervals, for example, collecting data every minute. In the data collected each time, the radar echo intensity distribution data is presented in the form of a two-dimensional matrix, and each element in the matrix represents the radar echo intensity value at different spatial positions. This intensity value reflects the degree of reflection of radar waves by precipitation particles in the convective cloud.

[0015] The convective cloud spatial distribution data combines the position information of the radar monitoring stations, meteorological satellite data, etc. to determine the specific range and shape of the convective cloud in space, and is also represented by a data structure, such as a set of polygon vertex coordinates. After the data at multiple time steps are collected in sequence, an extrapolation dataset of convective cloud radar echoes including radar echo intensity distribution data and corresponding convective cloud spatial distribution data at multiple time steps is formed. Then, the data collected from each radar monitoring station is summarized and sorted to ensure the accuracy and integrity of the data.

[0016] Step 120: Perform sparse sampling and spatio-temporal alignment processing on the extrapolation dataset of convective cloud radar echoes to generate a spatio-temporally continuous radar echo extrapolation reconstruction dataset.

[0017] In the embodiments of the present application, the obtained extrapolation dataset of convective cloud radar echoes is further sparsely sampled and spatio-temporally aligned to generate a spatio-temporally continuous radar echo extrapolation reconstruction dataset. As an optional embodiment, the sparse sampling and spatio-temporal alignment processing of the convective cloud radar echo extrapolation dataset to generate a spatio-temporally continuous radar echo extrapolation reconstruction dataset includes the following steps.

[0018] Step 121: Perform time alignment processing on the radar echo intensity distribution data of multiple time steps in the convective cloud radar echo extrapolation dataset according to a preset time resolution threshold to obtain a radar echo intensity distribution sequence with uniform time intervals.

[0019] In this meteorological monitoring scenario, the preset time resolution threshold is determined according to actual requirements and data analysis accuracy. For example, it is set to five minutes. When processing the radar echo intensity distribution data of multiple time steps, each time step data in the dataset is traversed. For cases where the time interval does not meet five minutes, a time interpolation algorithm, such as a linear interpolation algorithm, is used. For example, the data acquisition time intervals of two adjacent time steps are four minutes and six minutes. The radar echo intensity distribution data at the five-minute time point is calculated through linear interpolation, that is, the intensity values of these two adjacent time steps are weighted according to the time ratio to obtain the intensity value at the intermediate time point. By analogy, all time step data is processed, and finally a radar echo intensity distribution sequence with a uniform time interval of five minutes is obtained. This sequence provides a unified time scale for subsequent analysis, enabling data at different time steps to be compared and processed within the same time framework.

[0020] Step 122: Perform spatial interpolation processing on each radar echo intensity distribution data in the radar echo intensity distribution sequence with uniform time intervals to eliminate data missing areas in the radar detection blind area and obtain spatially continuous radar echo intensity distribution data.

[0021] In the embodiments of the present application, due to the blind area in radar monitoring, there will be some missing data in the data set. For each data in the radar echo intensity distribution sequence with uniform time intervals, that is, the two-dimensional radar echo intensity distribution matrix at each time step. For the data missing areas in the matrix, a spatial interpolation algorithm is adopted, such as the Kriging interpolation algorithm. This algorithm takes into account the spatial correlation of the data and comprehensively calculates factors such as the distance and direction of the surrounding known data points. For example, for a data missing point, find the known data points within a certain range around it, assign different weights according to the distance between these data points and the missing point, the closer the distance, the greater the weight, and then calculate the radar echo intensity value of the missing point through weighted average. Such processing is performed on all missing areas in the entire matrix, thereby eliminating the data missing areas in the radar detection blind area and obtaining spatially continuous radar echo intensity distribution data, which enables the radar echo intensity distribution of convective clouds in space to be presented completely, providing more accurate spatial information for subsequent analysis.

[0022] Step 123: Determine the spatial sampling interval based on the moving speed threshold of the convective cloud, and perform sparse sampling on the spatially continuous radar echo intensity distribution data according to the spatial sampling interval to generate the radar echo intensity distribution data after sparse sampling.

[0023] In the meteorological scenario of the embodiments of the present application, first, through the analysis of historical convective cloud data and the currently monitored convective cloud characteristics, a moving speed threshold of a convective cloud is determined, for example, set to 50 kilometers per hour. According to this moving speed threshold, combined with the spatial scale and analysis accuracy requirements, the spatial sampling interval is calculated. For example, after calculation, the spatial sampling interval is 1 kilometer. Then, the spatially continuous radar echo intensity distribution data is processed, and data points are selected in the two-dimensional radar echo intensity distribution matrix according to the spatial sampling interval of 1 kilometer. That is, the intensity value at a position is selected every 1 kilometer, and the data points at other positions are discarded to generate the radar echo intensity distribution data after sparse sampling. Such sparse sampling not only reduces the data volume and the computational complexity of subsequent processing but also retains the key information of the radar echo intensity distribution of convective clouds, enabling the data to be analyzed more efficiently in the subsequent process.

[0024] Step 124: Perform spatial superposition processing on the radar echo intensity distribution data after sparse sampling and the convective cloud spatial distribution data to generate a spatio-temporally continuous radar echo extrapolation and reconstruction data set.

[0025] In this meteorological monitoring area scenario, the radar echo intensity distribution data after sparse sampling, that is, a two-dimensional matrix containing intensity values at specific spatial positions, is spatially superimposed with the convective cloud spatial distribution data, such as the convective cloud spatial range data represented by a set of polygon vertex coordinates. By matching and corresponding each position in the radar echo intensity distribution data with the spatial range in the convective cloud spatial distribution data. For example, for one position in the radar echo intensity distribution matrix, if this position is within the polygon range of the convective cloud spatial distribution, then the intensity value at this position is associated with the convective cloud spatial information. By performing such processing on the entire data, a spatio-temporally continuous radar echo extrapolation and reconstruction data set is generated. This data set integrates data with uniformly spaced time intervals, continuous space, and superimposed convective cloud spatial information, providing a comprehensive and accurate data basis for subsequent convective cloud analysis.

[0026] Step 130: Call the spatio-temporal sequence prediction model to perform cyclic radar echo extrapolation processing on the radar echo extrapolation and reconstruction data set, generating an extrapolation result within a preset time range. The extrapolation result includes the spatial evolution path and intensity change trend of the convective cloud.

[0027] In the embodiment of the present application, the trained spatio-temporal sequence prediction model is used to operate on the spatio-temporally continuous radar echo extrapolation and reconstruction data set obtained through processing. This model learns and analyzes the features of the time and space dimensions in the data set to predict the radar echo situation of the convective cloud within a preset future time range, thereby obtaining an extrapolation result including the spatial evolution path and intensity change trend of the convective cloud.

[0028] As another optional embodiment, the calling the spatio-temporal sequence prediction model to perform cyclic radar echo extrapolation processing on the radar echo extrapolation and reconstruction data set, generating an extrapolation result within a preset time range, includes: Step 131: Input the spatio-temporally continuous radar echo extrapolation and reconstruction data set into the encoder module of the spatio-temporal sequence prediction model to generate spatio-temporal encoding features of the radar echo intensity distribution data.

[0029] In this meteorological monitoring scenario, a spatio-temporal continuous radar echo extrapolation and reconstruction dataset, that is, the radar echo intensity distribution and convective cloud spatial distribution data integrating time and space information, is input into the encoder module of the spatio-temporal sequence prediction model. Inside the encoder module, there are corresponding neural network layers and algorithms. For example, it may include convolutional layers, which extract features of the spatial information in the dataset through convolutional operations, associating and integrating the radar echo intensity values at different positions with the surrounding spatial information. At the same time, it may also include recurrent neural network layers to process the data in the time dimension, capturing the changing trends and patterns in the time series. Through the collaborative work of these layers, feature extraction and transformation are performed on the input dataset, converting it into spatio-temporal encoding features of the radar echo intensity distribution data. These features are a highly abstract data representation, containing the key information of the data in time and space, providing a basis for subsequent decoding and extrapolation.

[0030] Step 132: Perform time-step recursive decoding processing on the spatio-temporal encoding features through the decoder module of the spatio-temporal sequence prediction model to generate radar echo intensity distribution prediction data for each time step within a preset time range.

[0031] Next, the decoder module of the spatio-temporal sequence prediction model processes the spatio-temporal encoding features generated by the encoder module. The decoder module also contains a neural network structure and algorithms. It uses a time-step recursive method to start from the current moment and gradually predict the radar echo intensity distribution for each time step within a preset future time range, either forward or backward. For example, leveraging the recursive characteristics of the recurrent neural network, based on the spatio-temporal encoding features of the current time step and the prediction result of the previous time step, through corresponding calculations and transformations, the radar echo intensity distribution prediction data for the next time step is generated. During this process, the decoder module continuously interprets and transforms the spatio-temporal encoding features, converting the abstract features into specific radar echo intensity distribution prediction values at each time step, thus obtaining the prediction data for each time step within the preset time range.

[0032] Step 133: Perform spatial continuity constraint processing on the radar echo intensity distribution prediction data for each time step within the preset time range to generate spatially smooth radar echo intensity distribution prediction data.

[0033] After obtaining the predicted data of the radar echo intensity distribution at each time step within the preset time range, in order to make these data more reasonable and continuous in space, spatial continuity constraint processing is performed. This processing uses corresponding algorithms and rules to operate on the two-dimensional radar echo intensity distribution prediction data matrix at each time step. For example, using a spatial smoothing algorithm, for each position in the matrix, considering the intensity values of its surrounding adjacent positions, the intensity value of this position is adjusted by means of weighted average or the like. If the intensity value of one position differs greatly from the intensity values of its surrounding adjacent positions, the algorithm will correct it according to the surrounding values, making the intensity distribution in the entire space smoother and more continuous. Through such processing, spatially smoothed predicted data of the radar echo intensity distribution are generated, improving the credibility and usability of the predicted data in space.

[0034] Step 134: Determine the spatial evolution path and intensity change trend of the convective cloud based on the superposition result of the spatially smoothed predicted data of the radar echo intensity distribution and the convective cloud spatial distribution data, and generate the extrapolation result.

[0035] Finally, the spatially smoothed predicted data of the radar echo intensity distribution obtained through spatial continuity constraint processing are superimposed on the convective cloud spatial distribution data. During the superposition process, the predicted data of the radar echo intensity distribution at each time step are associated and matched with the corresponding convective cloud spatial range. For example, for one of the time steps, at different positions within the convective cloud spatial range, the intensity value of this position is determined according to the predicted data of the radar echo intensity distribution. By performing such superposition and analysis on the data of multiple time steps, observe the changes in the spatial position and intensity of the convective cloud at different time steps. Thus, determine the spatial evolution path of the convective cloud, that is, the trajectory passed by the convective cloud in space over time, and the intensity change trend, such as whether the intensity increases or decreases, etc., and finally generate an extrapolation result containing this information.

[0036] In an alternative embodiment, a specific method and process for training the spatio-temporal sequence prediction model are provided. Through a series of operations and algorithms, the model can learn the characteristics and laws of the convective cloud radar echo data, so as to have the ability to accurately predict. Based on this, the training method of the spatio-temporal sequence prediction model includes: Step 210: Obtain the historical convective cloud radar echo extrapolation data set and the corresponding true extrapolation result data, perform spatio-temporal alignment and noise filtering processing on the historical convective cloud radar echo extrapolation data set, and generate a training data set.

[0037] In the meteorological research scenario of the embodiment of the present application, first, a large number of historical convective cloud radar echo extrapolation datasets and the corresponding real extrapolation result data are collected. These historical data come from the monitoring records of convective clouds in different past time periods. Then, the historical convective cloud radar echo extrapolation datasets are processed. For the time dimension, a time alignment algorithm similar to that in step 121 is adopted. According to the set time resolution threshold, for example, set to ten minutes, the data at different time steps in the dataset are aligned to ensure the consistency of the data in time. For the spatial dimension, some spatial calibration and matching operations may also be involved to enable the accurate correspondence of data from different sources in space. At the same time, noise filtering processing is performed on the data using a filtering algorithm, such as the Gaussian filtering algorithm. For the noise in the dataset, such as outliers caused by radar equipment errors or external interference, the data is smoothed through this algorithm to remove the influence of noise. After these spatio-temporal alignment and noise filtering processes, a training dataset for training the spatio-temporal sequence prediction model is generated. This training dataset has the characteristics of high-quality and consistent data, providing a reliable basis for the training of the model.

[0038] Step 220: Build an initial spatio-temporal sequence prediction model, where the initial spatio-temporal sequence prediction model includes a spatio-temporal convolutional encoder, a time recurrent decoder, and a spatial continuity constraint module.

[0039] In this embodiment, the construction of the initial spatio-temporal sequence prediction model begins. The spatio-temporal convolutional encoder part consists of multiple convolutional layers. The parameters such as the convolutional kernel size and stride of these convolutional layers are set according to the data characteristics and model requirements. For example, convolutional kernels of different sizes are set. Small convolutional kernels are used to capture local spatial features, and large convolutional kernels are used to obtain more extensive spatial information. Through convolutional operations, feature extraction is performed on the input data in the spatial dimension, and the spatial information is transformed into a more abstract feature representation. The time recurrent decoder adopts a recurrent neural network structure, such as the long short-term memory network (LSTM) or the gated recurrent unit (GRU). These structures can effectively process time series data and learn and process the information in the time dimension through recursion to predict future situations based on past information. The spatial continuity constraint module contains some algorithms and rules for ensuring spatial smoothness and continuity, such as the spatial smoothing algorithm mentioned above. These three modules are combined together according to the set connection relationship to form the overall architecture of the initial spatio-temporal sequence prediction model, providing a basic framework for subsequent training and prediction.

[0040] Step 230: Input the training dataset into the initial spatio-temporal sequence prediction model, extract the multi-scale spatio-temporal features of the radar echo intensity distribution training data in the training dataset through the spatio-temporal convolutional encoder, and generate predicted radar echo intensity distribution data based on the multi-scale spatio-temporal features through the time recurrent decoder.

[0041] Optionally, the generated training data set is input into the established initial spatio-temporal sequence prediction model, and the spatio-temporal convolutional encoder starts to work. It processes the radar echo intensity distribution training data in the training data set. Through convolutional operations with different-sized convolutional kernels, spatio-temporal features of the data are extracted from different scales. For example, smaller convolutional kernels can capture the subtle change features of the radar echo intensity in local areas, while larger convolutional kernels can obtain more macroscopic spatial distribution features. At the same time, by combining the information in the time dimension, multi-scale spatio-temporal features are extracted. Then, the time recurrent decoder works based on these multi-scale spatio-temporal features. It utilizes the recursive characteristics of the recurrent neural network and, according to the already extracted spatio-temporal features, gradually generates the predicted radar echo intensity distribution data. For example, starting from the features of the current time step, combining the prediction results and spatio-temporal features of the previous time step, the radar echo intensity distribution of the next time step is predicted, thus generating the entire predicted data sequence.

[0042] Step 240: Invoke the spatial continuity constraint module to perform spatial smoothing constraint on the predicted radar echo intensity distribution data, and calculate the spatio-temporal consistency loss between the predicted radar echo intensity distribution data and the true extrapolation result data.

[0043] After the predicted radar echo intensity distribution data is generated, the spatial continuity constraint module is invoked to process it. The spatial continuity constraint module uses corresponding algorithms, such as the spatial smoothing algorithm mentioned above, to perform spatial smoothing on the predicted radar echo intensity distribution data at each time step to ensure the continuity and rationality of the data in space. Then, the spatio-temporal consistency loss between the predicted radar echo intensity distribution data and the true extrapolation result data is calculated. This loss calculation uses a mean square error loss function or a cross-entropy loss function, etc. By comparing the differences between the predicted data and the true data in the time and space dimensions, a value is calculated to represent the degree of inconsistency between the two. For example, for the predicted values and true values at each time step and spatial position, the differences between them are calculated, and then these differences are comprehensively calculated through the loss function to obtain the spatio-temporal consistency loss value. This loss value reflects the deviation degree of the model prediction result from the true situation and provides a basis for subsequent model parameter optimization.

[0044] Step 250: Optimize the parameters of the initial spatio-temporal sequence prediction model based on the spatio-temporal consistency loss until the spatio-temporal consistency loss converges, and obtain the trained spatio-temporal sequence prediction model.

[0045] Optionally, according to the calculated spatio-temporal consistency loss, the parameters of the initial spatio-temporal sequence prediction model are optimized. Optimization algorithms are adopted, such as the stochastic gradient descent algorithm or the adaptive moment estimation (Adam) algorithm, etc. These algorithms adjust the parameters in the model, such as the weights and biases of the neural network layers, according to the magnitude and change direction of the loss value. For example, by calculating the gradient of the loss function with respect to the parameters, and adjusting the parameter values according to the direction and magnitude of the gradient, the loss value is gradually reduced. This process is continuously repeated, that is, continuously inputting training data, calculating the loss, and optimizing the parameters until the spatio-temporal consistency loss converges. When the loss value no longer decreases significantly and reaches a relatively stable state, it indicates that the model has learned the features and patterns in the data. At this time, the trained spatio-temporal sequence prediction model is obtained, and this trained model can more accurately predict and analyze the convective cloud radar echo data.

[0046] Step 140: Generate a convective cloud rain attenuation operation strategy based on the extrapolation result, and send the convective cloud rain attenuation operation strategy to the rain attenuation operation service system for job processing. The convective cloud rain attenuation operation strategy includes the job point location, job timing, and job status evaluation features.

[0047] In the embodiment of the present application, based on the previously obtained convective cloud extrapolation result, through a series of analyses and calculations, a convective cloud rain attenuation operation strategy is formulated. This strategy covers key information such as the job point location, job timing, and job status evaluation features, and then this strategy is sent to a dedicated rain attenuation operation service system to guide the actual job execution.

[0048] In a preferred embodiment, the specific steps and operation methods for generating a convective cloud rain attenuation operation strategy based on the extrapolation result are further elaborated in detail, providing more specific guidance for practical applications. Based on this, the generating of the convective cloud rain attenuation operation strategy based on the extrapolation result includes: Step 141: Extract the spatial evolution path and intensity change trend of the convective cloud from the extrapolation result, and determine the moving direction and coverage area of the convective cloud within a preset time range.

[0049] In this meteorological operation scenario, from the obtained extrapolation results, carefully analyze the spatial evolution path and intensity change trend of convective clouds. By analyzing the spatial position information and intensity values of convective clouds at different time steps, determine the moving direction and coverage area of convective clouds within a preset time range. For example, by observing the coordinate changes of the core area of convective clouds in the extrapolation results at different time steps, calculate the direction vector of its movement to determine the moving direction. For the coverage area, by analyzing the spatial range data of convective clouds at each time step, determine the spatial range occupied by it at different times, and synthesize the data of multiple time steps to obtain the coverage area within the preset time range. This coverage area may be an irregular polygon area. Through a series of algorithms and data processing, accurately define its boundary and range, providing an important basis for determining the location and timing of operation points in the subsequent steps.

[0050] Step 142: Generate the operation timing according to the change trend of radar echo intensity within the coverage area, where the operation timing includes the operation start time and the operation duration.

[0051] After determining the coverage area of convective clouds, further analyze the change trend of radar echo intensity within this area. Observe the change of radar echo intensity at different positions within the coverage area in the extrapolation results over time. For example, it is found that the radar echo intensity shows a rapid upward trend during a certain period, and after the intensity value reaches a certain threshold, it may indicate that the development of convective clouds enters a critical stage. According to this intensity change trend, combined with the objectives and requirements of rain attenuation operations, determine the operation start time. For example, when the radar echo intensity rises to a point close to the peak and is about to start falling, set it as the operation start time to maximize the effect of rain attenuation operations. For the operation duration, determine it by analyzing factors such as the slope of the intensity change trend and the moving speed of convective clouds. If the intensity decreases relatively slowly and the moving speed of convective clouds is moderate, a longer operation duration may be set to ensure continuous operation intervention during the process of convective clouds passing through the target area. Through such analysis and calculation, generate the operation timing including the accurate operation start time and operation duration.

[0052] Step 143: Calculate the location of the rain attenuation operation point based on the moving direction and the coverage area, where the operation point location includes the area with the largest gradient change and the preset interception point on the core moving path of convective clouds.

[0053] After knowing the moving direction and coverage area of the known convective cloud, the calculation of the rain attenuation operation point location begins. First, analyze the gradient change of the radar echo intensity within the coverage area. By performing differential calculations on the radar echo intensities at different positions within the coverage area, the intensity gradient is obtained. For example, for each position within the coverage area, calculate the intensity difference between it and the adjacent position to form an intensity gradient field. In this gradient field, find the areas with the largest gradient changes. These areas usually correspond to the positions where the internal physical processes of the convective cloud change more violently and are one of the key areas for rain attenuation operations. At the same time, according to the moving direction and the core moving path of the convective cloud, select preset interception points on the core moving path. For example, according to the moving direction vector of the convective cloud, at a certain distance in front of its moving path, considering factors such as the boundary and shape of the coverage area, determine the preset interception points. These preset interception points are for performing operation intervention in advance during the movement of the convective cloud to achieve a better rain attenuation effect. Combine the areas with the largest gradient changes and the preset interception points to determine the rain attenuation operation point location.

[0054] In one implementation, the specific operation steps for calculating the rain attenuation operation point location based on the moving direction and coverage area are further refined. Through more detailed calculation and analysis methods, the accuracy and effectiveness of the operation point location are ensured. Based on this, the calculation of the rain attenuation operation point location based on the moving direction and coverage area includes: Step 1431: Extract the centroid coordinate sequence of the core area of the convective cloud at consecutive time steps within a preset time window according to the spatial evolution path of the convective cloud, and calculate the moving direction vector and average moving speed based on the centroid coordinate sequence.

[0055] In this meteorological operation scenario, select a preset time window from the spatial evolution path data of the convective cloud, for example, select the data within the next hour. Within this time window, extract the centroid coordinates of the core area of the convective cloud at consecutive time steps. For the spatial range data of the convective cloud at each time step, calculate the centroid coordinates of its core area through the corresponding algorithm. For example, for the spatial range of the convective cloud represented by a polygon, calculate the weighted average of the vertex coordinates of the polygon to obtain the centroid coordinates. Arrange these centroid coordinates at consecutive time steps in sequence to form a centroid coordinate sequence. Then, calculate the moving direction vector based on this centroid coordinate sequence. By calculating the difference between the centroid coordinates of adjacent time steps, obtain the displacement vector within each time interval, and comprehensively average these displacement vectors to obtain the moving direction vector, which represents the average moving direction of the convective cloud during this time period. At the same time, calculate the average moving speed. By calculating the total sum of the distances between adjacent centroids in the centroid coordinate sequence and dividing it by the total duration of the time window, obtain the average moving speed of the convective cloud within this preset time window. These moving direction vectors and average moving speeds provide important kinematic parameters for subsequent determination of the operation point location.

[0056] Step 1432: Perform a spatial projection operation on the boundary coordinates of the covered area and the moving direction vector to determine the geometric center point coordinates of the downstream protection area, and generate a polar coordinate system with the geometric center point as the origin.

[0057] It can be understood that after obtaining the moving direction vector of the convective cloud, the boundary coordinates of the covered area are processed. Perform a spatial projection operation on the coordinate of each boundary point of the covered area and the moving direction vector. For example, for a point on the boundary, project the point onto the straight line where the moving direction vector is located through the vector projection formula to obtain the coordinates of the projection point. By performing projection operations on all boundary points, the range of the downstream protection area is determined. Then, calculate the geometric center point coordinates of the downstream protection area. For example, for a downstream protection area in a polygonal shape, obtain the geometric center point coordinates by calculating the average value of the vertex coordinates of the polygon. Establish a polar coordinate system with this geometric center point as the origin. In the polar coordinate system, the position of each point can be represented by the polar radius and the polar angle. Such a coordinate system is more convenient for subsequent analysis and calculation of the position of the operation point, and can more intuitively reflect the relationship between the point and the moving direction of the convective cloud and the protection area.

[0058] Step 1433: Generate a radial detection path along the moving direction vector in the polar coordinate system, calculate the time evolution step size on each path based on the average moving speed, and generate a spatio-temporally correlated grid detection point array.

[0059] After establishing the polar coordinate system, with the origin as the center, generate a radial detection path along the moving direction vector. These detection paths are evenly distributed in the space centered on the origin and cover the downstream protection area. Based on the previously calculated average moving speed, calculate the time evolution step size on each path. For example, according to the average moving speed and the preset time resolution, determine the position points corresponding to a certain time interval on each path. For example, if the average moving speed is 60 kilometers per hour and the preset time resolution is one minute, then on each path, calculate the corresponding moving distance every minute according to the moving speed, so as to determine the position point on the path at that moment. By performing such calculations on all radial detection paths, generate a spatio-temporally correlated grid detection point array. Each detection point in this array corresponds to a time and a spatial position, providing a detailed spatial and temporal framework for subsequent analysis of the radar echo intensity and determination of the operation point.

[0060] Step 1434: Perform a radar echo intensity gradient field analysis on each detection point in the grid detection point array, and extract the vertical integrated liquid water content change rate and the horizontal echo intensity gradient modulus length of each detection point within the preset time step.

[0061] Optionally, for each detection point in the generated gridded detection point array, a detailed analysis of the radar echo intensity gradient field is performed. Within a preset time step, for example, data within the next ten minutes is selected to analyze the change rate of the vertically integrated liquid water content at each detection point. By integrating the liquid water content data at different altitude levels and then calculating the change within the preset time step, the change rate of the vertically integrated liquid water content is obtained. At the same time, the magnitude of the horizontal echo intensity gradient is calculated. For the horizontal space around each detection point, the gradient of the radar echo intensity is calculated and then its magnitude is taken to represent the degree of change of the radar echo intensity in the horizontal direction at that point. By performing such an analysis on each detection point, two important characteristic parameters of that point are obtained, which reflect the physical properties and change trends of the convective cloud at that position and provide a basis for subsequent determination of candidate operation points.

[0062] Step 1435: Perform dimensionless normalization on the change rate of the vertically integrated liquid water content and the magnitude of the horizontal echo intensity gradient to generate a composite influence factor parameter, and screen out the detection points whose composite influence factor parameter exceeds the target threshold as candidate operation points.

[0063] To comprehensively consider the influence of the change rate of the vertically integrated liquid water content and the magnitude of the horizontal echo intensity gradient on the selection of operation points, dimensionless normalization is performed on them. Using a normalization algorithm, for example, the change rate of the vertically integrated liquid water content and the magnitude of the horizontal echo intensity gradient are respectively subtracted from their respective means and then divided by their respective standard deviations to transform these two parameters into the same dimension range. Then, the two normalized parameters are combined in a certain form, such as weighted summation, to generate a composite influence factor parameter. A target threshold is set, for example, 0.8, and the detection points whose composite influence factor parameter exceeds this target threshold are screened out. These detection points have relatively significant characteristics in terms of the physical properties and changes of the convective cloud and are considered possible candidate operation points, providing a candidate set for subsequent further determination of the operation point location.

[0064] Step 1436: Generate a density clustering model based on the spatial distribution density of the candidate operation points, extract the cluster center coordinates through the density clustering model, and perform position offset compensation according to the geometric topological characteristics of the downstream protection area to generate the position of the rain reduction operation point.

[0065] After obtaining the set of candidate operation points, a density clustering model is generated based on the spatial distribution density of these candidate operation points. The density clustering algorithm is adopted, such as the DBSCAN algorithm. This algorithm divides the points with connected density into different clusters according to the distribution density of the candidate operation points in space. Through this algorithm, the dense areas and sparse areas in the candidate operation points can be discovered. Then, the coordinates of the cluster centers are extracted from each cluster, and these cluster centers represent the concentrated positions of the operation points in the cluster. However, due to the geometric topological features of the downstream protection area may be relatively complex, for example, there are irregular shapes or some special areas that need to be protected with emphasis, so it is necessary to perform position offset compensation on the coordinates of the cluster centers according to these geometric topological features. For example, if the cluster center is close to the boundary of the protection area and may exceed the effective operation range, the coordinates of the cluster center are appropriately adjusted according to the shape and requirements of the protection area to make it in a more suitable operation position. After such processing, the positions of the rain reduction operation points are finally generated.

[0066] Step 1437: Perform a correlation check on the projection components of the movement direction vectors of each operation point in the positions of the rain reduction operation points and the coverage area parameter, and output the optimized operation point position coordinate sequence after removing the spatially redundant points.

[0067] The positions of the generated rain reduction operation points are further optimized. Calculate the projection component of the movement direction vector of each operation point, that is, project the movement direction vector of the operation point onto one of the reference directions of the coverage area to obtain the projection component. At the same time, consider the coverage area parameter, such as the total area of the coverage area or the effective operation area, etc. By analyzing the correlation between the projection component of the movement direction vector of the operation point and the coverage area parameter, it is judged whether the spatial distribution of the operation points is reasonable. If the relationship between the projection component of one operation point and the coverage area shows that it may be redundant in spatial distribution, for example, the influence of this operation point on the coverage area is small and it overlaps with the functions of other operation points, it will be removed. After such correlation check and redundant point removal operations, the optimized operation point position coordinate sequence is output, and these coordinates represent the more reasonable and effective positions of the rain reduction operation points.

[0068] Step 144: Generate operation state evaluation features according to the area with the largest gradient change and the preset interception points, and the operation state evaluation features include the operation coverage matching degree and the operation effect persistence index.

[0069] After determining the area with the largest gradient change and the preset interception points, operation state evaluation features are generated based on this, and these features are used to evaluate the effect and state of the rain reduction operation during the implementation process, providing a basis for subsequent operation adjustment and optimization.

[0070] In another implementation, the specific operations and calculation methods for generating job status evaluation features based on the region with the largest gradient change and the preset interception point are described in detail to more accurately evaluate the job status. Based on this, generating the job status evaluation features according to the region with the largest gradient change and the preset interception point includes: Step 1441: Extract the spatio-temporal evolution parameters of the region with the largest gradient change from the extrapolation result, including the time series of the echo intensity gradient modulus length, the spatial integral value of the liquid water content, and the curvature radius of the core movement path.

[0071] From the previously obtained extrapolation result, detailed parameter extraction is performed for the region with the largest gradient change. First, extract the time series of the echo intensity gradient modulus length. At different time steps, calculate the radar echo intensity gradient within the region with the largest gradient change and take its modulus length to form a sequence that changes with time. This sequence reflects the evolution of the severity of the radar echo intensity change within this region over time. At the same time, calculate the spatial integral value of the liquid water content. Integrate the liquid water content at different altitude levels in the region with the largest gradient change to obtain a value that comprehensively reflects the total amount of liquid water in this region. The change in this value can reflect the change in water vapor content inside the convective cloud. In addition, extract the curvature radius of the core movement path. For the core movement path of the convective cloud, calculate its curvature radius through the corresponding algorithm. The size of the curvature radius reflects the degree of curvature of the convective cloud movement path, which is of great significance for analyzing the movement trend and stability of the convective cloud. These spatio-temporal evolution parameters describe the characteristics of the region with the largest gradient change from different aspects and provide basic data for generating job status evaluation features later.

[0072] Step 1442: Construct a three-dimensional spherical coordinate system within the spatial range where the preset interception point is located, and calculate the cumulative value of the echo intensity flux within a preset distance upstream of the interception point and the decay coefficient of the dissipation rate within a preset distance downstream.

[0073] Construct a three-dimensional spherical coordinate system within the spatial range where the preset interception point is located to more accurately analyze the distribution of physical quantities in this region. Taking the preset interception point as the center of the sphere, set a sphere space with a set radius range. Under this three-dimensional spherical coordinate system, calculate the cumulative value of the echo intensity flux within a preset distance upstream of the interception point. By integrating the radar echo intensities at different positions within the preset distance upstream and considering factors such as spatial position and direction, obtain the cumulative value of the echo intensity flux, which reflects the total amount of radar echo energy entering this region upstream of the interception point. At the same time, calculate the decay coefficient of the dissipation rate within a preset distance downstream. Observe the dissipation of the radar echo intensity over time within the preset distance downstream. Through the analysis and calculation of the intensity values at different time steps, obtain the decay coefficient of the dissipation rate, which represents the degree of attenuation of the radar echo intensity over time in the downstream region and reflects the influence effect of the rain attenuation operation on the convective cloud.

[0074] Step 1443: Perform time-phase matching processing on the time series of the magnitude of the echo intensity gradient and the dissipation rate attenuation coefficient to generate a time-varying correlation matrix and extract the principal component eigenvectors.

[0075] Optionally, perform time-phase matching processing on the extracted time series of the magnitude of the echo intensity gradient and the dissipation rate attenuation coefficient. Since the rhythms of these two parameters changing with time may be different, through the time-phase matching algorithm, they can be better aligned and compared in time. For example, adopt the phase synchronization algorithm to adjust the time axes of the two series so that their key features can correspond in time. Then, generate a time-varying correlation matrix based on the matched series. The elements of this matrix represent the degree of correlation between the two parameters at different time points. By calculating the correlation indexes of the two series at different time points, the matrix elements are filled. Next, perform principal component analysis on the time-varying correlation matrix to extract the principal component eigenvectors. The principal component eigenvectors can retain the information in the matrix to the greatest extent, reflecting the main change patterns and correlation features between the time series of the magnitude of the echo intensity gradient and the dissipation rate attenuation coefficient, and providing key information for generating the job status evaluation index in the subsequent steps.

[0076] Step 1444: Perform a dimensionless fusion operation on the spatial integral value of the liquid water content and the cumulative value of the echo intensity flux to obtain the job energy intervention effectiveness index, and calculate the path offset correction factor based on the curvature radius of the core movement path.

[0077] Optionally, perform a dimensionless fusion operation on the spatial integral value of the liquid water content and the cumulative value of the echo intensity flux. Adopt a fusion algorithm. For example, standardize the two parameters separately to make them within the same dimension range, and then fuse them together through weighted summation or other appropriate operation methods to obtain the job energy intervention effectiveness index. This index comprehensively reflects the intervention effect of the rain attenuation operation on the internal energy and water vapor content of the convective cloud. The higher the index, the greater the impact of the operation on the convective cloud and the better the effect. At the same time, calculate the path offset correction factor based on the curvature radius of the core movement path. For example, determine the correction factor according to the size and change of the curvature radius. If the curvature radius is small, it indicates that the movement path of the convective cloud is highly curved, and a larger correction factor may be required to adjust the operation strategy to adapt to the movement changes of the convective cloud. This path offset correction factor is used to correct the job status evaluation, taking into account the influence of the movement path of the convective cloud.

[0078] Step 1445: Construct a multi-dimensional state space according to the principal component eigenvectors, and map the job energy intervention effectiveness index and the path offset correction factor into the state space to generate a state trajectory curve.

[0079] Optionally, a multi-dimensional state space is constructed based on the extracted principal component feature vectors. The dimension of the principal component feature vectors determines the dimension of the state space. For example, if there are two principal component feature vectors, a two-dimensional state space is constructed. The operation energy intervention efficiency index and the path deviation correction factor are used as the coordinate values in the state space respectively, and they are mapped into this multi-dimensional state space. As time goes by, these two parameters will change, forming a trajectory curve in the state space, that is, the state trajectory curve. This curve reflects the comprehensive state of the operation at different time points. By analyzing the curve, the evolution of the operation effect over time can be understood, as well as the comprehensive influence of the operation energy intervention and the convective cloud movement path on the operation state.

[0080] Step 1446: Extract the time parameters and spatial coordinates corresponding to the curvature extreme points of the state trajectory curve, and generate a coverage matching degree evaluation matrix in combination with the operation time window of the preset interception point.

[0081] Optionally, analyze the state trajectory curve, and extract the time parameters and spatial coordinates corresponding to its curvature extreme points. The curvature extreme points represent the points where the change trend of the state trajectory curve changes significantly, and the time and spatial coordinates corresponding to these points are of great significance. In combination with the operation time window of the preset interception point, these time parameters and spatial coordinates are associated and matched with the operation time window. For example, judge whether the curvature extreme points fall within the operation time window and their specific positions within the window. According to this information, a coverage matching degree evaluation matrix is generated. The elements of the matrix represent the matching degree between the operation coverage range and the actual effect in different situations. For example, the rows of the matrix can represent different time points, and the columns can represent different spatial regions. The element values are assigned according to the results of the association and matching, so as to comprehensively evaluate the relationship between the operation coverage range and the actual operation effect.

[0082] Step 1447: Perform singular value decomposition on the coverage matching degree evaluation matrix, extract the eigenvector corresponding to the largest singular value as the operation coverage range matching degree, and calculate the stability index of the eigenvector evolving over time as the operation effect persistence parameter.

[0083] In this step, singular value decomposition is performed on the generated coverage matching degree evaluation matrix. Singular value decomposition is a matrix decomposition method that can decompose a matrix into the product of three matrices, where the singular values reflect the important characteristics of the matrix. Through singular value decomposition, the eigenvector corresponding to the largest singular value is extracted. This eigenvector synthesizes the key information in the matrix and is used as an index for the matching degree of the operation coverage range. It reflects the best matching relationship between the operation coverage range and the actual effect. At the same time, the stability index of the evolution of this eigenvector over time is calculated. By observing the changes of the eigenvector at different time points and calculating some statistical indicators of the eigenvector, such as variance, etc., to measure its stability. This stability index is used as a parameter for the persistence of the operation effect, reflecting the duration of the operation effect over time. The higher the stability, the longer the operation effect lasts and the more stable the effect is.

[0084] Step 1448: Linearly weighted synthesize the operation coverage range matching degree and the operation effect persistence parameter to generate an operation status evaluation eigenvector in a comprehensive quantitative form.

[0085] After obtaining the operation coverage range matching degree and the operation effect persistence parameter, in order to comprehensively evaluate the operation status, these two parameters are linearly weighted synthesized. According to actual requirements and experience, corresponding weights are assigned to the operation coverage range matching degree and the operation effect persistence parameter respectively. For example, weight w1 is assigned to the operation coverage range matching degree, and weight w2 is assigned to the operation effect persistence parameter, and w1 + w2 = 1. Through linear weighted calculation, that is, each dimension value of the operation status evaluation eigenvector is equal to the operation coverage range matching degree multiplied by w1 plus the operation effect persistence parameter multiplied by w2, thus generating an operation status evaluation eigenvector in a comprehensive quantitative form. This vector comprehensively reflects the comprehensive performance of the operation in terms of coverage range and effect persistence, providing a comprehensive and quantitative index for evaluating the status of the convective cloud rain attenuation operation, facilitating subsequent analysis and comparison of the operation effect, and adjusting and optimizing the operation strategy according to the evaluation results.

[0086] Step 145: Integrate the operation timing, operation point location, and operation status evaluation features to generate the convective cloud rain attenuation operation strategy.

[0087] After separately determining the operation timing, operation point location, and operation status evaluation features, these key elements are integrated to form a complete convective cloud rain attenuation operation strategy. Specifically, the operation timing information including the operation start time and operation duration, the sequence of operation point location coordinates obtained through calculation and optimization, and the comprehensively quantified operation status evaluation feature vector are organically combined. These information are organized together through a set data structure or format. For example, a structure or data table containing the information of these three parts can be constructed (in actual implementation, the appropriate data structure is determined according to the specific programming environment and requirements). The generated convective cloud rain attenuation operation strategy comprehensively covers the key aspects such as the time and location of operation implementation and the evaluation of operation effects, providing clear and detailed operation guidance for the rain attenuation operation service system, ensuring that the operation can be carried out at the appropriate time and location, and enabling effective evaluation and monitoring of the operation effects, so as to timely adjust the operation strategy and improve the efficiency and effect of the rain attenuation operation.

[0088] In an exemplary embodiment, it is elaborated in detail how to convert the generated convective cloud rain attenuation operation strategy into specific instructions and send them to the rain attenuation operation service system to implement the actual operation execution process. Based on this, the sending of the convective cloud rain attenuation operation strategy to the rain attenuation operation service system for operation processing includes: Step 146: Generate a geographic coordinate instruction according to the operation point location in the convective cloud rain attenuation operation strategy. The geographic coordinate instruction includes longitude and latitude information and the operation height range; convert the operation timing into a time control instruction for the operation equipment. The time control instruction includes the operation start moment, operation duration, and operation interval period; generate an operation resource scheduling instruction according to the operation status evaluation feature. The resource scheduling instruction includes the catalyst dosage and equipment deployment priority; generate an operation instruction set based on the geographic coordinate instruction, time control instruction, and resource scheduling instruction, and send the operation instruction set to the rain attenuation operation service system to execute multi-device collaborative operation.

[0089] First, process the operation point location information in the convective cloud rain attenuation operation strategy. The operation point location is represented in a certain coordinate form. Through a coordinate conversion algorithm, it is converted into geographic coordinates, that is, longitude and latitude information. At the same time, according to the operation requirements and the actual situation of the convective cloud, determine the operation height range of each operation point. For example, the operation height range of one operation point may be from 1000 meters to 3000 meters above the ground. Combine the longitude and latitude information and the operation height range to form a geographic coordinate instruction.

[0090] Next, convert the operation timing. Information such as the operation start time and operation duration is converted into a time control instruction that the operation device can recognize and execute according to the control requirements of the operation device. For example, convert the operation start time into the absolute start time point of the device, and convert the operation duration into the running duration setting of the device. In addition, considering the situation where multiple operation devices may cooperate in operation, it is also necessary to determine the operation interval period, that is, the time interval between two adjacent operations. Integrate this information to form the time control instruction of the operation device.

[0091] Then, generate an operation resource scheduling instruction according to the operation status evaluation characteristics. The operation status evaluation characteristics reflect the expected effects and requirements of the operation. Based on this information, determine the amount of catalyst to be put at each operation point. For example, if the operation point is in an area where convective clouds are developing vigorously, more catalyst may need to be put. At the same time, determine the device deployment priority according to factors such as the importance of the operation and the performance of the device. For example, for critical operation points, devices with better performance are preferentially deployed. Combine information such as the amount of catalyst to be put and the device deployment priority to form the operation resource scheduling instruction.

[0092] Finally, integrate the geographic coordinate instruction, time control instruction, and resource scheduling instruction to generate an operation instruction set. This set contains all the key information required for operation execution. Send the operation instruction set to the rain attenuation operation service system through network communication or other appropriate means. After receiving the instruction set, the rain attenuation operation service system coordinates multiple operation devices according to the information in it to achieve multi-device cooperative operation. For example, deploy the device to the specified location according to the geographic coordinate instruction, control the start and running time of the device according to the time control instruction, and allocate the amount of catalyst to be put and arrange the device deployment order according to the resource scheduling instruction, so as to ensure that the rain attenuation operation can be accurately and efficiently executed according to the predetermined strategy.

[0093] As an optional embodiment, the method further includes: Step 300: Obtain the operation execution data fed back by the rain attenuation operation service system in real time. The operation execution data includes the actual operation point location and the radar echo monitoring data after the operation; perform a matching degree analysis on the actual operation point location and the operation point location in the convective cloud rain attenuation operation strategy to generate a position deviation correction feature; generate an effect deviation correction feature according to the difference between the radar echo monitoring data after the operation and the extrapolation result; perform online parameter adjustment on the spatio-temporal sequence prediction model based on the position deviation correction feature and the effect deviation correction feature.

[0094] This alternative embodiment describes how to adjust and optimize the spatio-temporal sequence prediction model according to actual feedback data during the execution of an operation, so as to improve the accuracy of the model and the operation effect. During the execution of the rain attenuation operation, the operation execution data fed back by the rain attenuation operation service system is obtained in real time. The actual operation point location is the location information where the operation equipment actually arrives and executes the operation. By comparing and analyzing it with the operation point location in the operation strategy, the matching degree between the two is calculated. For example, a distance metric algorithm is used to calculate the Euclidean distance or other appropriate distance metric value between the actual operation point and the planned operation point, and then it is judged whether the matching degree meets the requirements according to a preset threshold. If there is a deviation between the actual operation point location and the planned operation point location, a location deviation correction feature is generated according to information such as the magnitude and direction of the deviation.

[0095] At the same time, the radar echo monitoring data after the operation is obtained, and these data reflect the actual state of the convective cloud after the operation. The radar echo monitoring data after the operation is compared and analyzed with the extrapolation result generated by the previous spatio-temporal sequence prediction model. The differences between the two in terms of radar echo intensity distribution, convective cloud spatial range, etc. are compared. For example, the difference in radar echo intensity at different positions is calculated, and the change in the convective cloud spatial range is statistically analyzed, etc. According to these differences, an effect deviation correction feature is generated, and these features reflect the gap between the actual operation effect and the expected effect.

[0096] Finally, based on the location deviation correction feature and the effect deviation correction feature, online parameter adjustment of the spatio-temporal sequence prediction model is performed. An online learning algorithm is used, such as the stochastic gradient descent online learning algorithm, to adjust the parameters of the model according to the deviation correction feature. If the location deviation is large, the model parameters related to spatial positioning may be adjusted; if the effect deviation is obvious, the parameters related to radar echo intensity prediction, convective cloud evolution simulation, etc. are adjusted. By continuously adjusting the model parameters according to the actual feedback data, the spatio-temporal sequence prediction model can more accurately reflect the actual situation of the convective cloud, improve the prediction accuracy of the model, and further optimize the subsequent convective cloud rain attenuation operation strategy.

[0097] As a non-limiting embodiment, after the convective cloud rain attenuation operation strategy is sent to the rain attenuation operation service system for operation processing, the method further includes: Step 400: Obtain the spatial trajectory data of the operation equipment and the measured data of the liquid water content of the meteorological monitoring station in real time. Decompose the spatial trajectory data into moving direction vectors to generate equipment movement trajectory characteristic parameters; perform time-domain difference operation based on the measured data of the liquid water content and the predicted liquid water content data in the extrapolation result to generate the influence factor correction coefficient for each operation point; calculate the coverage overlap rate between adjacent operation points according to the equipment movement trajectory characteristic parameters, and generate a target weight matrix in combination with the influence factor correction coefficient; perform density resampling on the operation points in the uncovered area through the target weight matrix, generate a set of derived operation point coordinates and update them to the rain reduction operation service system.

[0098] This embodiment describes how to further optimize and adjust the operation strategy according to the spatial trajectory data of the operation equipment and the measured data of the liquid water content of the meteorological monitoring station after the operation is executed, so as to improve the coverage effect and pertinence of the operation. Specifically, the spatial trajectory data of the operation equipment during the operation process is collected in real time, and these data record the movement path of the operation equipment in space. Decompose the spatial trajectory data into moving direction vectors. By analyzing the position changes of the equipment at different times, calculate the moving direction vector and speed of the equipment and other parameters to generate equipment movement trajectory characteristic parameters, which can accurately describe the movement state of the operation equipment and provide a basis for subsequent analysis of the operation coverage.

[0099] At the same time, obtain the measured data of the liquid water content of the meteorological monitoring station, and perform time-domain difference operation on these measured data and the predicted liquid water content data in the extrapolation result of the previous spatio-temporal sequence prediction model. For each operation point, calculate the difference between the measured liquid water content and the predicted liquid water content at different time steps, and then generate the influence factor correction coefficient for each operation point according to these differences. This coefficient reflects the difference degree between the actual liquid water content and the predicted value and can be used to correct the operation strategy.

[0100] According to the equipment movement trajectory characteristic parameters, calculate the coverage overlap rate between adjacent operation points. By analyzing the movement trajectory of the operation equipment, determine whether there is overlap in the operation coverage range of adjacent operation points and the degree of overlap. For example, calculate the ratio of the intersection area to the union area of the coverage areas of two operation points to obtain the coverage overlap rate. Combine the influence factor correction coefficient, and generate a target weight matrix according to the importance and actual effect of different operation points. The elements in the matrix represent the weights of each operation point, and the weight size is determined comprehensively according to the coverage overlap rate and the influence factor correction coefficient.

[0101] Finally, density resampling is performed on the job points in the uncovered area using the target weight matrix. For areas that are not fully covered, the sampling probability of each job point is determined according to the weight matrix. The higher the weight of a job point, the higher the probability of being sampled. Through the resampling algorithm, a set of coordinates of derivative job points is generated. These new job point coordinates are generated based on the actual operation situation and data feedback, and are more targeted and reasonable. The set of coordinates of derivative job points is updated to the rain abatement operation service system, enabling the operation system to adjust operations according to the new job points, improving the coverage effect and overall efficiency of the operations.

[0102] As a non-limiting embodiment, after sending the convective cloud rain abatement operation strategy to the rain abatement operation service system for operation processing, it further includes: Step 500: Extract the three-dimensional wind field vector data of each height layer in the operation area, perform a vector superposition operation on the wind field vector data and the spatial evolution path in the extrapolation result to generate a corrected movement path; calculate the catalyst diffusion rate compensation factor according to the curvature parameter of the corrected movement path, and generate a stratified delivery rate control instruction in combination with the operation height range; fit the parabola equation to the catalyst sedimentation trajectory of each job point, and adjust the equation coefficients based on the diffusion rate compensation factor to generate a three-dimensional diffusion model; based on the stratified delivery rate control instruction and the three-dimensional diffusion model, adjust the catalyst release pulse frequency and injection elevation angle in real time.

[0103] This embodiment elaborates on how to use the three-dimensional wind field vector data and extrapolation results in the operation area to finely control the delivery of the catalyst after the operation is executed, so as to improve the diffusion effect of the catalyst and the operation efficiency. First, extract the three-dimensional wind field vector data of different height layers in the operation area. These data reflect the wind direction and wind speed information at different heights in the operation area. Perform a vector superposition operation on the wind field vector data and the spatial evolution path of the convective cloud in the extrapolation result of the spatio-temporal sequence prediction model. For example, for the spatial position and movement direction of the convective cloud at a certain moment, add the wind field vector at that position to obtain a corrected movement path. This corrected movement path takes into account the influence of the wind field on the movement of the convective cloud and more accurately reflects the actual movement trajectory of the convective cloud.

[0104] Calculate the catalyst diffusion rate compensation factor according to the curvature parameter of the corrected movement path. The curvature parameter reflects the degree of bending of the corrected movement path. The greater the degree of bending, the greater the possible influence on the diffusion of the catalyst. Further, a compensation factor is calculated according to the curvature magnitude to adjust the diffusion rate of the catalyst. In combination with the operation height range, the meteorological conditions and catalyst diffusion requirements are different at different height layers. Determine the stratified delivery rate of the catalyst for each height layer according to the height range, and generate a stratified delivery rate control instruction.

[0105] The catalyst sedimentation trajectory of each operation point is fitted with a parabola equation. By collecting the data of catalyst sedimentation in actual operations and using curve fitting algorithms such as the least squares method, a parabola equation is fitted to describe the catalyst sedimentation trajectory. Based on the diffusion rate compensation factor calculated previously, the coefficients of the parabola equation are adjusted to generate a three-dimensional diffusion model that can more accurately reflect the diffusion of the catalyst under the current meteorological conditions.

[0106] Finally, based on the layered dosing rate control instruction and the three-dimensional diffusion model, the catalyst release pulse frequency and injection elevation angle are adjusted in real time. According to the layered dosing rate control instruction, the catalyst dosing amounts at different height layers and different time points are determined, and the dosing amount is controlled by adjusting the catalyst release pulse frequency. At the same time, according to the three-dimensional diffusion model and the actual operation requirements, the injection elevation angle is adjusted to enable the catalyst to diffuse at a suitable position and direction, improving the utilization efficiency of the catalyst and the effect of the rain suppression operation, and ensuring efficient and accurate catalyst dosing under different meteorological conditions and convective cloud states.

[0107] As a non-limiting embodiment, after sending the convective cloud rain suppression operation strategy to the rain suppression operation service system for job processing, it further includes: Step 600: Obtain the mapping relationship table between the Doppler radar reflectivity factor of the operation area and the catalytic dose dosing, and find the optimal catalytic dose gradient in the mapping relationship table based on the real-time reflectivity change rate; calculate the resource utilization rate index according to the remaining catalyst reserve of the operation equipment and the optimal catalytic dose gradient, and generate the resource allocation status characteristics of each operation equipment; perform a moving average process on the resource allocation status characteristics within a preset time window to generate a target resource scheduling sequence; perform dosing amount attenuation control on non-critical operation points based on the target resource scheduling sequence, and allocate an excess dosing task to critical operation points.

[0108] The embodiment of the present application illustrates how to reasonably schedule the catalyst dosing according to the mapping relationship between the Doppler radar reflectivity factor of the operation area and the catalytic dose dosing, and the resource situation of the operation equipment after the operation is executed, so as to optimize the resource utilization and operation effect. First, obtain the mapping relationship table between the Doppler radar reflectivity factor of the operation area and the catalytic dose dosing. This table is established through a large number of experiments and historical data, and it records the corresponding optimal catalytic dose under different reflectivity factor conditions. Based on the real-time monitored Doppler radar reflectivity change rate, find the corresponding optimal catalytic dose gradient in the mapping relationship table. For example, if the reflectivity change rate is relatively fast, it indicates that the development of the convective cloud is relatively rapid, and according to the mapping relationship table, a larger catalytic dose gradient is determined at this time.

[0109] Calculate the resource utilization rate index based on the remaining catalyst reserve of the operation equipment and the optimal catalyst dosage gradient. For example, by calculating the ratio of the remaining catalyst reserve of each operation equipment to the required catalyst reserve calculated according to the optimal catalyst dosage gradient, the resource utilization rate index is obtained. This index reflects the resource utilization situation of each operation equipment, generates the resource allocation status characteristics of each operation equipment, and these characteristics can be represented by data vectors or other suitable data structures, comprehensively describing the resource status of each operation equipment.

[0110] Within a preset time window, for example, set to ten minutes, perform a moving average process on the resource allocation status characteristics. By continuously updating the data within the time window, calculate the average value of the resource allocation status characteristics, and generate the target resource scheduling sequence, which reflects the average situation and trend of the resource allocation of the operation equipment over a period of time.

[0111] Finally, classify the operation points and perform resource scheduling based on the target resource scheduling sequence. Determine the critical operation points and non-critical operation points according to the importance and effect evaluation of the operations. For non-critical operation points, perform decay control on the dosing amount according to the target resource scheduling sequence and the resource situation, reducing the dosing amount of the catalyst. For critical operation points, in order to ensure the operation effect, allocate an over-dose task and give priority to ensuring the catalyst supply of critical operation points. Through such a resource scheduling method, optimize the dosing of the catalyst, improve the resource utilization efficiency, and ensure the best rain attenuation operation effect under limited resource conditions.

[0112] It can be understood that after completing the above operations such as formulating and implementing the convective cloud rain attenuation operation strategy, in order to further improve the effect and efficiency of the convective cloud rain attenuation operation, it can also be implemented in combination with the following embodiments.

[0113] First, when using the radar base data of a single-station radar for many years in the data preparation link, use the python radar library wradlib to read the data, and perform appropriate quality control on all elevation echoes through a texture filter to ensure the data quality. Subsequently, interpolate the data into a Cartesian grid with a spatial resolution of using the nearest four points to form a combined radar reflectivity product. For the convenience of storage and subsequent processing, use the maximum radar detection distance of 230 km to store the data as a single-channel grayscale image in PNG format, and set the image resolution to 。Next, all severe convection processes are screened out from these data to construct a severe convection radar echo extrapolation dataset. The screening criteria are set to ensure that the maximum echo value of each sample (20 frames, 2h) is above 50dBZ and the maximum echo value of each frame is above 30dBZ. Severe convection processes that meet the requirements are screened out through the above strict criteria. Continuous sampling is adopted, that is, a radar map is taken every 6 minutes, and the window size is set to 20 frames (input 10 frames and predict 10 frames). The dataset is constructed in a sliding window manner to provide sufficient and appropriate data support for subsequent model training.

[0114] In addition, during the model selection and optimization process, multiple mainstream spatio-temporal sequence prediction models are used, and the key parameters of the models are set respectively, such as the number of stacked layers, the number of hidden neurons, the learning rate, etc. These mainstream models include ConvLSTM, PredRNN, PredRNN++, MIM, PredRNN-V2, SimVP-V1, SimVP-V2, TAU, etc. Based on the constructed severe convection radar echo extrapolation dataset, these models are trained and optimized, and the optimal model is selected by comparing the performance of each model. For example, different models have their own advantages and disadvantages in capturing the spatio-temporal characteristics of convective clouds and prediction accuracy. After a large number of experiments and evaluations, the model most suitable for this dataset and task is determined. Further, the adjustment of the sampling method also has an important impact on the results. Considering that the emergence of the operation effect takes a long time and the over-decay characteristics of the model, in order to avoid the operation effect being masked and improve the extrapolation effect, the sampling method is modified. Sparse sampling is adopted, that is, a radar map is taken every 12 minutes to reconstruct the dataset. Based on the original dataset, a radar map is taken every other radar map to form a sparse sampling severe convection radar echo extrapolation dataset. This sampling method can better adapt to the operation requirements and model characteristics while reducing the data volume.

[0115] Furthermore, based on the sparse sampling severe convection radar echo extrapolation dataset, after setting the key parameters using the optimal model, training and optimization are carried out to obtain the weights of the optimal training results. Since the error of the model extrapolation result increases sharply with the increase of the time step, in order to avoid the operation effect being masked and relying on the model's learning of the normal development and dissipation speed of clouds, only the first frame of the extrapolation result, that is, the 12-minute extrapolation result, is extracted and extrapolated in a 12-minute cycle to obtain a longer extrapolation result. From the start of the operation, the optimal result weights obtained by training and optimizing based on the sparse sampling severe convection radar echo extrapolation dataset are used to perform a 12-minute cycle radar echo extrapolation that only extracts the first frame, that is, the 12-minute extrapolation result. This extrapolation result can be used as a reference for the normal development and dissipation speed of clouds.

[0116] During the implementation of the operation, it is very important to determine the analysis target area. Due to the uncertainty of the affected area of cloud seeding operations, obvious changes may also occur around the direct affected area of the operation. The embodiments of this application only focus on the changes in the direct affected area of the operation. Based on the location of the operation point and the moving direction and speed of the cloud, the diffusion range of silver iodide is speculated, and the direct affected area of the rain reduction operation is framed as the analysis target area, providing a clear range for subsequent effect evaluation.

[0117] In the operation effect evaluation link, the combined radar reflectivity change evaluation analysis of the rain reduction operation effect in the direct affected area every 12 minutes can be carried out using the difference between the extrapolated radar map and the actual radar map. In this way, qualitatively judge the weakening position of the cloud radar reflectivity, and comprehensively judge the development trend change of the cloud by qualitatively judging the entire movement process of the cloud within the direct affected area of the operation. At the same time, using the recommended Z-R relationship of summer convective cloud precipitation in the United States in the extrapolated radar map, calculate the rainfall of the cloud within the direct affected area during the operation process, and compare it with the actual rainfall to quantitatively analyze the rain reduction effect of the rain reduction operation. Through the evaluation method combining qualitative and quantitative, the actual effect of the rain reduction operation can be comprehensively and accurately understood, providing a strong basis for the adjustment and optimization of subsequent operation strategies.

[0118] In addition, during the actual implementation process, the data missing in the radar blind area can be eliminated by the adaptive adjustment of the semi-variogram parameters in the Kriging interpolation algorithm, and the spatio-temporal alignment can be achieved by combining the dimensional unification processing of the convective cloud moving speed and the spatial resolution (for example, converting the speed threshold of 50 km / h to m / s and matching it with the spatial sampling interval); when constructing the spatio-temporal convolutional encoder, a multi-scale convolutional kernel combination (such as 3×3 and 5×5 convolutions in parallel) is used to extract local and global features, and the time recurrence mechanism of the gated recurrent unit (GRU) is used to capture the sequence dependence relationship; for the processing of spatial continuity constraints, the spatial regularization of the echo intensity distribution can be realized based on the dynamic generation of the neighborhood weight matrix in the Laplace smoothing algorithm; in the construction of the density clustering model, the optimal clustering center is determined by the adaptive calculation of the ε-neighborhood radius in the DBSCAN algorithm (based on the standard deviation of the spatial distribution of candidate operation points); for the fusion operation of the liquid water content and the echo intensity, the entropy method is used to determine the objective weight coefficient of the characteristic parameters to avoid subjective experience deviation; at the same time, the projection component of the moving path is corrected based on the three-dimensional wind field data assimilation technology in the WRF meteorological model, and the three-dimensional diffusion model is optimized by combining the catalyst diffusion rate and the Reynolds number correction factor of the parabolic equation.

[0119] In one aspect of the embodiments of the present application, by performing sparse sampling and spatio-temporal alignment processing on the convective cloud radar echo extrapolation data set of the target area, data can be effectively integrated to generate a spatio-temporally continuous radar echo extrapolation reconstruction data set, greatly improving the coherence and usability of the data, and enabling the radar echo extrapolation reconstruction data set to more accurately reflect the actual situation of convective clouds. In another aspect, by invoking a spatio-temporal sequence prediction model for cyclic radar echo extrapolation processing, potential patterns in the data can be deeply mined, and extrapolation results that include the spatial evolution path and intensity change trend of convective clouds within a preset time range can be accurately generated, providing a strong basis for subsequent decision-making. In yet another aspect, the convective cloud rain attenuation operation strategy generated based on the extrapolation results can comprehensively consider the operation point location, operation timing, and operation status evaluation characteristics, making the rain attenuation operation more targeted and scientific. Sending the convective cloud rain attenuation operation strategy to the rain attenuation operation service system for operation processing can effectively optimize the operation process, improve the accuracy and efficiency of the rain attenuation operation, and provide reliable support for meteorological operations.

[0120] In summary, the embodiments of the present application obtain a comprehensive convective cloud radar echo extrapolation data set, perform innovative data processing, accurately generate extrapolation results, and formulate reasonable operation strategies, effectively solving problems such as insufficient data utilization and unscientific operation strategy formulation in the prior art for convective cloud rain attenuation operations, and significantly improving the accuracy and effectiveness of operations.

[0121] Based on the same inventive concept, the embodiments of the present application also provide a convective cloud rain attenuation operation analysis system. Refer to Figure 2 As shown, it is a schematic structural diagram of a possible convective cloud rain attenuation operation analysis system provided in the embodiments of the present application. Figure 2 In this, the convective cloud rain attenuation operation analysis system 200 includes: a processor 210 and a memory 220. Among them, the memory 220 stores a computer program executable by the processor 210. By executing the instructions stored in the memory 220, the processor 210 can execute the steps of the above-mentioned convective cloud rain attenuation operation analysis method based on deep learning.

[0122] Based on the same inventive concept, embodiments of the present application provide a computer-readable storage medium, which includes a computer program. When the computer program runs on a convective cloud rain attenuation operation analysis system, the computer program is used to cause the convective cloud rain attenuation operation analysis system to execute the steps of the above-mentioned convective cloud rain attenuation operation analysis method based on deep learning. In some possible implementation manners, various aspects of the convective cloud rain attenuation operation analysis method provided by the present application can also be implemented in the form of a program product, which includes a computer program. When the program product runs on a convective cloud rain attenuation operation analysis system, the computer program is used to cause the convective cloud rain attenuation operation analysis system to execute the steps in the above-mentioned convective cloud rain attenuation operation analysis method based on deep learning. For example, the convective cloud rain attenuation operation analysis system can execute the steps as shown in Figure 1 shown in.

[0123] In the technical solutions involved in the above embodiments of the present invention, whether it is to perform comparison calculations of multi-dimensional features or construct composite parameters, if there are problems caused by significant differences in the number of dimensions, dimension units, and semantic meanings of different features, those skilled in the art, based on their professional knowledge and past practical experience, can fully understand that these differences need to be properly handled to make the calculation results accurate and comparable, and avoid situations such as logical confusion and unclear mathematical meanings.

[0124] Specifically, when facing features with different numbers of dimensions, in order to accurately calculate the similarity, matching degree, or feature distance between different features, those skilled in the art can use a variety of strategies, including but not limited to feature selection, feature extraction, and kernel function processing.

[0125] When processing the comparison of multi-dimensional features, in order to achieve comparable alignment of the feature space, those skilled in the art can adopt a variety of existing general technical means, including but not limited to standardization preprocessing, mapping transformation, and space projection.

[0126] In the process of constructing a composite parameter (such as a loss function value), different parameter items often have different dimensions. Those skilled in the art can adopt normalization processing or an adaptive weight allocation mechanism based on distribution characteristics.

[0127] The above general technical means for solving the problems of feature matching and loss balance all belong to the common general knowledge in the art. These technical means have been fully verified and widely used in a large number of practical applications. Those skilled in the art can proficiently and flexibly use these methods to handle similar dimension difference problems.

[0128] In the embodiments of the present application, the formulas and calculation processes involved, whether used for multi-dimensional feature comparison or composite loss function construction, strictly follow the principle of dimension correspondence. Each variable in the formula has a clear and definite physical meaning, and its operation logic fully conforms to the basic mathematical and physical logics. The operation result is necessarily a reasonable result expected by the present application. Those skilled in the art are capable of comprehensively applying the above general technical means according to specific data situations and business requirements to effectively solve various problems brought about by the number of dimensions, dimension differences, etc. in the multi-dimensional feature comparison calculation and composite loss function construction in the embodiments, ensuring the accuracy, reliability, and feasibility of the technical solution of the present invention.

Claims

1. A method for analyzing the rain reduction operation of convective clouds based on deep learning, characterized in that, Including: Obtain the extrapolation dataset of convective cloud radar echo in the target area, where the extrapolation dataset of convective cloud radar echo includes radar echo intensity distribution data of multiple time steps and corresponding convective cloud spatial distribution data; Perform sparse sampling and spatio-temporal alignment processing on the extrapolation dataset of convective cloud radar echo to generate a spatio-temporally continuous radar echo extrapolation reconstruction dataset; Call a spatio-temporal sequence prediction model to perform cyclic radar echo extrapolation processing on the radar echo extrapolation reconstruction dataset to generate extrapolation results within a preset time range, where the extrapolation results include the spatial evolution path and intensity change trend of convective clouds; Generate a convective cloud rain attenuation operation strategy based on the extrapolation results, and send the convective cloud rain attenuation operation strategy to the rain attenuation operation service system for job processing. The convective cloud rain attenuation operation strategy includes the location of the operation point, the operation timing, and the operation status evaluation characteristics.

2. The method according to claim 1, characterized in that, The performing sparse sampling and spatio-temporal alignment processing on the extrapolation dataset of convective cloud radar echo to generate a spatio-temporally continuous radar echo extrapolation reconstruction dataset includes: Perform time alignment processing on the radar echo intensity distribution data of multiple time steps in the extrapolation dataset of convective cloud radar echo according to a preset time resolution threshold to obtain a radar echo intensity distribution sequence with uniform time intervals; Perform spatial interpolation processing on each radar echo intensity distribution data in the radar echo intensity distribution sequence with uniform time intervals to eliminate data missing areas in the radar detection blind area, and obtain spatially continuous radar echo intensity distribution data; Determine the spatial sampling interval based on the moving speed threshold of convective clouds, and perform sparse sampling on the spatially continuous radar echo intensity distribution data according to the spatial sampling interval to generate sparsely sampled radar echo intensity distribution data; Perform spatial superposition processing on the sparsely sampled radar echo intensity distribution data and the convective cloud spatial distribution data to generate a spatio-temporally continuous radar echo extrapolation reconstruction dataset.

3. The method according to claim 1, wherein The calling a spatio-temporal sequence prediction model to perform cyclic radar echo extrapolation processing on the radar echo extrapolation reconstruction dataset to generate extrapolation results within a preset time range includes: Input the spatio-temporally continuous radar echo extrapolation reconstruction dataset into the encoder module of the spatio-temporal sequence prediction model to generate spatio-temporal encoding features of the radar echo intensity distribution data; Perform time-step recursive decoding processing on the spatio-temporal encoding features through the decoder module of the spatio-temporal sequence prediction model to generate radar echo intensity distribution prediction data for each time step within a preset time range; Perform spatial continuity constraint processing on the radar echo intensity distribution prediction data for each time step within the preset time range to generate spatially smooth radar echo intensity distribution prediction data; Determine the spatial evolution path and intensity change trend of convective clouds according to the superposition result of the spatially smooth radar echo intensity distribution prediction data and the convective cloud spatial distribution data, and generate the extrapolation results.

4. The method according to claim 3, characterized in that The training method of the spatio-temporal sequence prediction model includes: Obtain the historical extrapolation dataset of convective cloud radar echoes and the corresponding true extrapolation result data, and perform spatio-temporal alignment and noise filtering on the historical extrapolation dataset of convective cloud radar echoes to generate a training dataset; Build an initial spatio-temporal sequence prediction model, which includes a spatio-temporal convolutional encoder, a time-recurrent decoder, and a spatial continuity constraint module; Input the training dataset into the initial spatio-temporal sequence prediction model, extract the multi-scale spatio-temporal features of the radar echo intensity distribution training data in the training dataset through the spatio-temporal convolutional encoder, and generate predicted radar echo intensity distribution data based on the multi-scale spatio-temporal features through the time-recurrent decoder; Call the spatial continuity constraint module to perform spatial smoothing constraint on the predicted radar echo intensity distribution data, and calculate the spatio-temporal consistency loss between the predicted radar echo intensity distribution data and the true extrapolation result data; Optimize the parameters of the initial spatio-temporal sequence prediction model based on the spatio-temporal consistency loss until the spatio-temporal consistency loss converges to obtain a trained spatio-temporal sequence prediction model.

5. The method according to claim 1, characterized in that, The generation of the convective cloud rain reduction operation strategy based on the extrapolation result includes: Extract the spatial evolution path and intensity change trend of the convective cloud from the extrapolation result, and determine the moving direction and coverage area of the convective cloud within a preset time range; Generate the operation timing according to the radar echo intensity change trend within the coverage area, and the operation timing includes the operation start time and the operation duration; Calculate the positions of the rain reduction operation points based on the moving direction and the coverage area, and the positions of the operation points include the area with the largest gradient change and the preset interception points on the core moving path of the convective cloud; Generate operation status evaluation features according to the area with the largest gradient change and the preset interception points, and the operation status evaluation features include the operation coverage range matching degree and the operation effect persistence index; Integrate the operation timing, the positions of the operation points, and the operation status evaluation features to generate the convective cloud rain reduction operation strategy.

6. The method according to claim 5, wherein The calculation of the positions of the rain reduction operation points based on the moving direction and the coverage area includes: Extract the centroid coordinate sequence of the core area of the convective cloud at consecutive time steps within a preset time window according to the spatial evolution path of the convective cloud, and calculate the moving direction vector and the average moving speed based on the centroid coordinate sequence; Perform a spatial projection operation on the boundary coordinates of the coverage area and the moving direction vector to determine the geometric center point coordinates of the downstream protection area, and generate a polar coordinate system with the geometric center point as the origin; Generate a radial detection path along the moving direction vector in the polar coordinate system, calculate the time evolution step length on each path based on the average moving speed, and generate a spatio-temporally correlated grid detection point array; Perform radar echo intensity gradient field analysis on each detection point in the grid detection point array, and extract the vertical integrated liquid water content change rate and the horizontal echo intensity gradient modulus length of each detection point within a preset time step; Perform dimensionless standardization processing on the vertical integrated liquid water content change rate and the magnitude of the horizontal echo intensity gradient to generate a composite influence factor parameter, and screen the detection points where the composite influence factor parameter exceeds the target threshold as candidate operation points; Generate a density clustering model based on the spatial distribution density of the candidate operation points, extract the cluster center coordinates through the density clustering model, and perform position offset compensation according to the geometric topological characteristics of the downstream protection area to generate the position of the rain reduction operation points; The method further includes: Perform a correlation check on the projection components of the movement direction vectors of the operation points in the position of the rain reduction operation points and the coverage area parameter, and output the optimized operation point position coordinate sequence after removing spatially redundant points.

7. The method according to claim 5, characterized in that, The generating the operation state evaluation feature according to the region with the largest gradient change and the preset interception point includes: Extract the spatio-temporal evolution parameters of the region with the largest gradient change from the extrapolation result, including the time series of the magnitude of the echo intensity gradient, the spatial integral value of the liquid water content, and the curvature radius of the core movement path; Construct a three-dimensional spherical coordinate system within the spatial range where the preset interception point is located, and calculate the cumulative value of the echo intensity flux within a preset distance upstream of the interception point and the decay coefficient of the dissipation rate within a preset distance downstream; Perform time-phase matching processing on the time series of the magnitude of the echo intensity gradient and the decay coefficient of the dissipation rate to generate a time-varying correlation matrix and extract the principal component eigenvectors; Perform dimensionless fusion operation on the spatial integral value of the liquid water content and the cumulative value of the echo intensity flux to obtain the operation energy intervention efficiency index, and calculate the path offset correction factor based on the curvature radius of the core movement path; Construct a multi-dimensional state space according to the principal component eigenvectors, and map the operation energy intervention efficiency index and the path offset correction factor into the state space to generate a state trajectory curve; Extract the time parameters and spatial coordinates corresponding to the curvature extreme points of the state trajectory curve, and generate a coverage matching degree evaluation matrix in combination with the operation time window of the preset interception point; Perform singular value decomposition on the coverage matching degree evaluation matrix, extract the eigenvector corresponding to the largest singular value as the operation coverage matching degree, and calculate the stability index of the eigenvector evolving with time as the operation effect persistence parameter; Perform linear weighted synthesis on the operation coverage matching degree and the operation effect persistence parameter to generate an operation state evaluation feature vector in a comprehensive quantitative form.

8. The method according to claim 1, wherein The method further includes: Obtain in real time the operation execution data feedback by the rain reduction operation service system, where the operation execution data includes the actual operation point position and the radar echo monitoring data after the operation; Perform a matching degree analysis on the actual operation point position and the operation point position in the convective cloud rain reduction operation strategy to generate a position deviation correction feature; Generate an effect deviation correction feature according to the difference between the radar echo monitoring data after the operation and the extrapolation result; Perform online parameters on the spatio-temporal sequence prediction model based on the position deviation correction feature and the effect deviation correction feature.

9. The method according to claim 1, characterized in that, The sending the convective cloud rain reduction operation strategy to the rain reduction operation service system for operation processing includes: Generate a geographic coordinate instruction according to the location of the operation point in the convective cloud rain reduction operation strategy, where the geographic coordinate instruction includes longitude and latitude information and an operation altitude range; Convert the operation timing into a time control instruction for the operation equipment, where the time control instruction includes an operation start time, an operation duration, and an operation interval period; Generate an operation resource scheduling instruction according to the operation status evaluation feature, where the resource scheduling instruction includes a catalyst dosage and a device deployment priority; Generate a set of operation instructions based on the geographic coordinate instruction, the time control instruction, and the resource scheduling instruction, and send the set of operation instructions to the rain reduction operation service system to perform multi-device collaborative operations.

10. A convective cloud rain reduction operation analysis system, characterized in that, It includes a processor and a memory. Among them, the memory stores a computer program. When the computer program is executed by the processor, the processor is caused to execute the steps of any one of the methods recited in claims 1 to 9.

Images