Boundary layer height diagnosis and prediction method and system based on COSMIC data

Abstract

The invention discloses a boundary layer height diagnosis and prediction method and system based on COSMIC data. The method comprises the steps that boundary layer height diagnosis is carried out based on the COSMIC data to obtain boundary layer height data of different positions; taking the boundary layer height data of different positions as labels, and collecting longitudes and latitudes of corresponding positions and temperature and humidity at a moment before a specified prediction step length to construct a training data set; and constructing a machine learning model, and using the training data set to train and establish a mapping relationship between the input latitude and longitude, temperature and humidity and the predicted boundary layer height value output by prediction, so as to predict the boundary layer height predicted value of the target position according to the latitude and longitude, temperature and humidity of any target position. The invention aims to realize the height of the boundary layer by using the COSMIC data, and can quickly and accurately obtain the height information of the boundary layer in a large range and predict the height of the boundary layer.

Description

Technical Field

[0001] The present invention relates to the technical field of weather and climate forecasting, and in particular to a boundary layer height diagnosis and prediction method and system based on COSMIC data. Background Art

[0002] Real and accurate planetary boundary layer information plays a vital role in weather and climate forecasting systems, especially the boundary layer height, which is the determining factor of the type and number of covered clouds, affects the earth's radiation balance, and is closely related to surface heat flux, pollutant diffusion and waveguide. At present, the main ways to obtain boundary layer height information are radiosonde, radar, satellite remote sensing, etc. The atmospheric information provided by radiosonde data and radar echoes is real-time and effective, and the boundary layer height can be determined by establishing appropriate mathematical mapping relationships. However, it is difficult to deploy radiosonde and radar detection equipment on the sea surface, and it is not easy to obtain atmospheric boundary layer information, making it difficult to study the sea boundary layer height over a large range. Relatively speaking, satellite remote sensing technology has a wide detection range, is cheap to obtain data, and can calculate the boundary layer height using cloud top temperature and sea surface temperature detected by satellite. However, this method is limited to cloud-covered areas, and many related methods proposed by scholars are limited to cloudy or clear sky conditions, and the results are not ideal.

[0003] The COSMIC occultation detection technology developed in recent years is not affected by clouds or weather, and can detect information on the global boundary layer height. The data it produces has the characteristics of high coverage and high resolution. In recent years, the methods for calculating the boundary layer height using the refractive index of COSMIC data can be roughly divided into two categories: the maximum gradient method and the refractive index breakpoint method. The maximum gradient method has a relatively simple calculation process and relatively accurate calculation results, but it is less sensitive to small disturbances at the top of the boundary layer, so it is difficult to find the top of the boundary layer with weak refractive index changes. The refractive index breakpoint method calculates the boundary layer height by finding the refractive index breakpoint, although the process is complicated and sensitive to refractive index disturbances. Summary of the invention

[0004] Technical problem to be solved by the present invention: In view of the above-mentioned problems in the prior art, a boundary layer height diagnosis and prediction method and system based on COSMIC data are provided. The present invention aims to realize the boundary layer height by using COSMIC data, and can quickly, accurately and widely obtain boundary layer height information and predict the boundary layer height.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: A boundary layer height diagnosis and prediction method based on COSMIC data comprises the following steps: S1, boundary layer height diagnosis based on COSMIC data to obtain boundary layer height data at different locations; S2, taking the boundary layer height data at different locations as labels, collecting the longitude and latitude of the corresponding location and the temperature and humidity at the specified prediction step before to construct a training data set; S3, build a machine learning model, use the training data set to train and establish a mapping relationship between the input longitude and latitude, temperature and humidity and the predicted output boundary layer height prediction value, so as to predict the boundary layer height prediction value of the target location according to the longitude and latitude, temperature and humidity of any target location.

[0006] Optionally, step S1 includes: S1.1, resample the refractive index data in COSMIC data; S1.2, calculating the refractive index gradient segment by segment based on the refractive index data after data resampling; S1.3, filtering the boundary layer tops according to the preset boundary layer top filtering rules, thereby obtaining boundary layer height data at different positions corresponding to each boundary layer top, wherein the preset boundary layer top filtering rules include: Rule (1): The boundary layer top must satisfy: ,in is the refractive index gradient corresponding to the top of the boundary layer, is the height of the top of the boundary layer, is the refractive index gradient, min is the minimum value; Rule (2): The refractive index gradient value corresponding to the top of the boundary layer Less than the preset threshold; Rule (3): The height of the top of the boundary layer Less than a preset threshold; Rule (4): The number of minimum points on a refractive index profile that satisfies rules (1) to (3) is less than a preset threshold; Rule (5): There are no two minimum points on the refractive index profile whose difference is less than their respective preset ratios, where the preset ratio is less than 1.

[0007] Optionally, data resampling in step S1.1 refers to using cubic spline interpolation to interpolate the refractive index data in the COSMIC data from the original vertical resolution to obtain a higher-precision vertical resolution.

[0008] Optionally, calculating the refractive index gradient in segments according to the refractive index data after data resampling in step S1.2 includes: using a sliding window of a specified size for the refractive index data after data resampling, fitting a fitting curve to the refractive index points in the sliding window, and calculating the refractive index gradient of each segment of the fitting curve according to the following formula: , In the above formula, is the fitting curve, is the refractive index gradient, z represents the vertical height, and B is a constant coefficient.

[0009] Optionally, the refractive index gradient value corresponding to the boundary layer top in rule (2) of step S1.3 is Less than the preset threshold means the refractive index gradient value corresponding to the top of the boundary layer satisfy: , the height of the boundary layer top in rule (3) Less than the preset threshold means , the number of minimum points of a refractive index profile satisfying rules (1) to (3) in rule (4) is less than a preset threshold value means that the number of minimum points of a refractive index profile satisfying rules (1) to (3) is less than 7, and the absence of two minimum points on the refractive index profile whose differences are less than their respective preset ratios in rule (5) means that the absence of two minimum points on the refractive index profile whose differences are less than 30% of their respective ratios.

[0010] Optionally, when filtering the boundary layer tops according to the preset boundary layer top filtering rule in step S1.3, it also includes calculating the sharpness coefficient S according to the following formula for the filtered boundary layer tops and deleting the boundary layer tops whose sharpness coefficient S is less than a preset threshold value for the filtered boundary layer tops:

[0011] In the above formula, is the refractive index gradient at the top of the selected boundary layer, The root mean square of the minimum value of the refractive index gradient within a specified height of the refractive index profile.

[0012] Optionally, the machine learning model constructed in step S3 is a long short-term memory network LSTM.

[0013] In addition, the present invention also provides a boundary layer height diagnosis prediction system based on COSMIC data, comprising a microprocessor and a memory connected to each other, wherein the microprocessor is programmed or configured to execute the boundary layer height diagnosis prediction method based on COSMIC data.

[0014] In addition, the present invention also provides a computer-readable storage medium, in which a computer program or instruction is stored. The computer program or instruction is programmed or configured to execute the boundary layer height diagnosis and prediction method based on COSMIC data through a processor.

[0015] In addition, the present invention also provides a computer program product, including a computer program or an instruction, wherein the computer program or the instruction is programmed or configured to execute the boundary layer height diagnosis and prediction method based on COSMIC data through a processor.

[0016] Compared with the prior art, the present invention mainly has the following advantages: although the prior art can diagnose the boundary height according to COSMIC, it cannot predict the boundary layer height of any subsequent target position. The present invention, based on the boundary layer height diagnosis based on COSMIC data to obtain boundary layer height data of different positions, uses the boundary layer height data of different positions as labels, collects the longitude and latitude of the corresponding position and the temperature and humidity at the time before the specified prediction step to construct a training data set, uses the training data set to train and establish a mapping relationship between the input longitude and latitude, temperature and humidity and the predicted output boundary layer height prediction value, so as to predict the boundary layer height prediction value of the target position according to the longitude and latitude and temperature and humidity of the arbitrary target position, and can realize the subsequent boundary layer height prediction of any target position, and can quickly, accurately and widely obtain boundary layer height information and predict the boundary layer height. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Schematic diagram of the basic flow of the method of the embodiment of the present invention.

[0018] Figure 2 The flowchart of the method according to the embodiment of the present invention includes the step of performing boundary layer height diagnosis.

[0019] Figure 3 Schematic diagram of the results obtained in the embodiments of the present invention. DETAILED DESCRIPTION

[0020] like Figure 1 As shown, the boundary layer height diagnosis and prediction method based on COSMIC data in this embodiment includes the following steps: S1, boundary layer height diagnosis based on COSMIC data to obtain boundary layer height data at different locations; S2, taking the boundary layer height data at different locations as labels, collecting the longitude and latitude of the corresponding location and the temperature and humidity at the specified prediction step before to construct a training data set; S3, build a machine learning model, use the training data set to train and establish a mapping relationship between the input longitude and latitude, temperature and humidity and the predicted output boundary layer height prediction value, so as to predict the boundary layer height prediction value of the target location according to the longitude and latitude, temperature and humidity of any target location.

[0021] The use of radiosonde and radar detection equipment to obtain boundary layer heights has harsh conditions for setting up equipment, and it is difficult to obtain boundary layer heights over a large range of seas. Satellite-borne remote sensing technology has a wide detection range, but is limited by weather conditions and has high requirements for cloud coverage. However, COSMIC occultation detection technology is not affected by weather and can detect global boundary layer information. The data it produces has the characteristics of high coverage and high resolution. Therefore, boundary layer height diagnosis based on COSMIC data to obtain boundary layer height data at different locations can obtain accurate and comprehensive data, thereby improving the accuracy of boundary layer height prediction.

[0022] The traditional method of obtaining boundary layer height data based on COSMIC data is to use the refractive index breakpoint method. Because it is sensitive to the disturbance of the refractive index, the boundary layer height calculated by diagnosis is different from the actual height and will affect the prediction result. As an optional implementation method, Figure 2 As shown, step S1 of this embodiment includes: S1.1, resample the refractive index data in COSMIC data; S1.2, calculating the refractive index gradient segment by segment based on the refractive index data after data resampling; S1.3, filtering the boundary layer tops according to the preset boundary layer top filtering rules, thereby obtaining boundary layer height data at different positions corresponding to each boundary layer top, wherein the preset boundary layer top filtering rules include: Rule (1): The boundary layer top must satisfy: ,in is the refractive index gradient corresponding to the top of the boundary layer, is the height of the top of the boundary layer, is the refractive index gradient, min is the minimum value; Rule (2): The refractive index gradient value corresponding to the top of the boundary layer Less than the preset threshold; Rule (3): The height of the top of the boundary layer Less than a preset threshold; Rule (4): The number of minimum points that satisfy rules (1) to (3) on a refractive index profile is less than a preset threshold; Rule (5): There are no two minimum points on the refractive index profile whose difference is less than their respective preset ratios, where the preset ratio is less than 1. The boundary layer top is selected according to the preset boundary layer top selection rule. Figure 2 In the step of "determining the boundary layer height", the boundary layer height data at different positions corresponding to each boundary layer top are obtained. Figure 2 Then, you can construct a training dataset and train a machine learning model including a deep learning algorithm based on the constructed training dataset to obtain the boundary layer height prediction value.

[0023] In step S1.1 of this embodiment, data resampling refers to using cubic spline interpolation to interpolate the refractive index data in the COSMIC data from the original vertical resolution to obtain a higher-precision vertical resolution. Specifically, in this embodiment, the vertical resolution of the refractive index data in the original COSMIC data is 100m, and the vertical resolution is converted to 20m using the cubic spline interpolation method.

[0024] In step S1.2 of this embodiment, the refractive index gradient is calculated in segments according to the refractive index data after data resampling, including: using a sliding window of a specified size for the refractive index data after data resampling, fitting a fitting curve to the refractive index points in the sliding window (in this embodiment, 400 m is used as a fixed sliding window, that is, a curve is fitted using 20 refractive index points in the window), and calculating the refractive index gradient of each segment of the fitting curve according to the following formula: , In the above formula, is the fitting curve, the unit is unit, which represents the change of refractive index with height; is the refractive index gradient (calculation object), the unit is unit / Km; z represents the vertical height, the unit is Km; B is a constant coefficient, and the value is 300 in this embodiment.

[0025] The relevant values ​​in rules (2) to (5) of step S1.3 can be set according to actual needs. For example, as an optional implementation, the refractive index gradient value corresponding to the top of the boundary layer in rule (2) of step S1.3 is Less than the preset threshold means the refractive index gradient value corresponding to the top of the boundary layer satisfy: , the height of the boundary layer top in rule (3) Less than the preset threshold means , the number of minimum points of a refractive index profile satisfying rules (1) to (3) in rule (4) is less than a preset threshold value means that the number of minimum points of a refractive index profile satisfying rules (1) to (3) is less than 7, and the absence of two minimum points on the refractive index profile whose differences are less than their respective preset ratios in rule (5) means that the absence of two minimum points on the refractive index profile whose differences are less than 30% of their respective ratios.

[0026] Since the machine learning model in this embodiment learns the mapping relationship between the input latitude and longitude, temperature and humidity and the predicted output boundary layer height prediction value, it is possible to consider whether the temperature and water vapor have changed dramatically to achieve the screening of the boundary layer top. Specifically, in step S1.3 of this embodiment, when screening the boundary layer top according to the preset boundary layer top screening rules, it also includes calculating the sharp coefficient S for the screened boundary layer top according to the following formula and deleting the boundary layer top whose sharp coefficient S is less than the preset threshold for the screened boundary layer top:

[0027] In the above formula, is the refractive index gradient at the top of the selected boundary layer, The root mean square of the minimum refractive index gradient within a specified height (e.g. 10 km) of the refractive index profile. The sharpness coefficient S represents the strength of the refractive index gradient. The larger the coefficient, the more drastic the changes in temperature and water vapor, and the more accurate the determination of the top of the boundary layer.

[0028] The machine learning model constructed in step S3 of this embodiment is a long short-term memory network LSTM (Long Short-term Memory). The long short-term memory network LSTM is a special type of recurrent neural network (Recurrent Neural Network, RNN) model. Through its unique design, it effectively solves the problems of long-term dependence and gradient explosion that may be encountered in general recurrent neural network and artificial neural network models, and is suitable for time series prediction problems with long time intervals. In this embodiment, after the boundary layer height is calculated using COSMIC data, the global boundary layer height is predicted based on the long short-term memory network LSTM. The process of constructing and training the long short-term memory network LSTM is as follows: 1) Data preprocessing: The boundary layer height data set obtained by the above diagnostic process is sorted into tabular data in chronological order, and normalized with characteristic dimensions such as temperature, humidity, and longitude and latitude to eliminate the dimension effect. 2) Define the model: According to the classic long short-term memory network LSTM already on the network, use the Python built-in library to define the model, including its number of layers, learning rate and other parameters, and set the prediction step length to 24 hours (the prediction step length can be set according to actual needs). 3) Model training and visualization: Because the amount of data used is large, it is trained in batches, and the best model parameters are saved and plotted. After using the training data set to train the LSTM network to establish the mapping relationship between the input latitude and longitude, temperature and humidity and the predicted output boundary layer height prediction value, the network parameters of the LSTM network are saved, and the LSTM network can be used to predict the boundary layer height prediction value of the target location based on the latitude and longitude, temperature and humidity of any target location.

[0029] In order to verify the boundary layer height diagnosis and prediction method based on COSMIC data in this embodiment, the boundary layer height prediction value of the target location obtained by long short-term memory network LSTM in this embodiment is compared with the boundary layer height of the target location 24 hours later obtained by boundary layer height diagnosis based on COSMIC data. The results are as follows: Figure 3 As shown, the red curve is the predicted value of the boundary layer height at the target location obtained by the long short-term memory network LSTM, and the blue curve is the boundary layer height at the target location 24 hours later obtained by boundary layer height diagnosis based on COSMIC data. Figure 3 It can be seen that the boundary layer height prediction value of the target location predicted by the long short-term memory network LSTM in this embodiment is basically the same as the boundary layer height fluctuation of the target location 24 hours later obtained by boundary layer height diagnosis based on COSMIC data, indicating that the long short-term memory network LSTM in this embodiment can accurately predict the boundary layer height prediction value of the target location.

[0030] In summary, considering that COSMIC occultation detection technology is not affected by weather and can detect global boundary layer information, the data it generates has the characteristics of high coverage and high resolution. The method of this embodiment uses COSMIC data to perform boundary layer height diagnosis to obtain boundary layer height data at different locations, which can obtain accurate and comprehensive data, thereby improving the accuracy of boundary layer height prediction. Moreover, the advanced "open-loop" tracking technology adopted by the method of this embodiment can significantly solve the problem of atmospheric multipath, which is conducive to obtaining a more accurate boundary layer height. The method of this embodiment has fewer formulas, and only needs to use data processing and conversion. The boundary layer top height can be determined according to the threshold of the judgment condition. It is simple and convenient, and the calculation speed is fast. In addition, the method of this embodiment uses a mature long short-term memory network, and the prediction accuracy is guaranteed.

[0031] In addition, this embodiment also provides a boundary layer height diagnosis prediction system based on COSMIC data, including a microprocessor and a memory connected to each other, wherein the microprocessor is programmed or configured to execute a boundary layer height diagnosis prediction method based on COSMIC data.

[0032] In addition, this embodiment also provides a computer-readable storage medium, in which a computer program or instruction is stored. The computer program or instruction is programmed or configured to execute a boundary layer height diagnosis and prediction method based on COSMIC data through a processor.

[0033] In addition, this embodiment also provides a computer program product, including a computer program or instructions, which are programmed or configured to execute a boundary layer height diagnosis and prediction method based on COSMIC data through a processor.

[0034] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0035] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A boundary layer height diagnosis and prediction method based on COSMIC data, characterized in that: The steps include: S1, boundary layer height diagnosis based on COSMIC data to obtain boundary layer height data at different locations; S2, taking the boundary layer height data at different locations as labels, collecting the longitude and latitude of the corresponding location and the temperature and humidity at the specified prediction step before to construct a training data set; S3, build a machine learning model, use the training data set to train and establish a mapping relationship between the input longitude and latitude, temperature and humidity and the predicted output boundary layer height prediction value, so as to predict the boundary layer height prediction value of the target location according to the longitude and latitude, temperature and humidity of any target location.

2. The boundary layer height diagnosis and prediction method based on COSMIC data according to claim 1 is characterized in that: Step S1 includes: S1.1, resample the refractive index data in COSMIC data; S1.2, calculating the refractive index gradient segment by segment based on the refractive index data after data resampling; S1.3, filtering the boundary layer tops according to the preset boundary layer top filtering rules, thereby obtaining boundary layer height data at different positions corresponding to each boundary layer top, wherein the preset boundary layer top filtering rules include: Rule (1): The boundary layer top must satisfy: ,in is the refractive index gradient corresponding to the top of the boundary layer, is the height of the top of the boundary layer, is the refractive index gradient, min is the minimum value; Rule (2): The refractive index gradient value corresponding to the top of the boundary layer Less than the preset threshold; Rule (3): The height of the top of the boundary layer Less than a preset threshold; Rule (4): The number of minimum points on a refractive index profile that satisfies rules (1) to (3) is less than a preset threshold; Rule (5): There are no two minimum points on the refractive index profile whose difference is less than their respective preset ratios, where the preset ratio is less than 1.

3. The boundary layer height diagnosis and prediction method based on COSMIC data according to claim 2 is characterized in that: Data resampling in step S1.1 refers to using cubic spline interpolation to interpolate the refractive index data in the COSMIC data from the original vertical resolution to a higher-precision vertical resolution.

4. The boundary layer height diagnosis and prediction method based on COSMIC data according to claim 2 is characterized in that: The step S1.2 of calculating the refractive index gradient in sections according to the refractive index data after data resampling includes: using a sliding window of a specified size for the refractive index data after data resampling, fitting a fitting curve to the refractive index points in the sliding window, and calculating the refractive index gradient of each fitting curve according to the following formula: ， In the above formula, is the fitting curve, is the refractive index gradient, z represents the vertical height, and B is a constant coefficient.

5. The boundary layer height diagnosis and prediction method based on COSMIC data according to claim 2, characterized in that: The refractive index gradient value corresponding to the boundary layer top in rule (2) of step S1.3 Less than the preset threshold means the refractive index gradient value corresponding to the top of the boundary layer satisfy: , the height of the boundary layer top in rule (3) Less than the preset threshold means , the number of minimum points of a refractive index profile satisfying rules (1) to (3) in rule (4) is less than a preset threshold value means that the number of minimum points of a refractive index profile satisfying rules (1) to (3) is less than 7, and the absence of two minimum points on the refractive index profile whose differences are less than their respective preset ratios in rule (5) means that the absence of two minimum points on the refractive index profile whose differences are less than 30% of their respective ratios.

6. The boundary layer height diagnosis and prediction method based on COSMIC data according to claim 2 is characterized in that: In step S1.3, when filtering the boundary layer tops according to the preset boundary layer top filtering rules, it also includes calculating the sharpness coefficient S according to the following formula for the filtered boundary layer tops and deleting the boundary layer tops whose sharpness coefficient S is less than the preset threshold value for the filtered boundary layer tops: In the above formula, is the refractive index gradient at the top of the selected boundary layer, The root mean square of the minimum value of the refractive index gradient within a specified height of the refractive index profile.

7. The boundary layer height diagnosis and prediction method based on COSMIC data according to claim 1 is characterized in that: The machine learning model constructed in step S3 is a long short-term memory network LSTM.

8. A boundary layer height diagnosis and prediction system based on COSMIC data, comprising a microprocessor and a memory connected to each other, characterized in that: The microprocessor is programmed or configured to execute the boundary layer height diagnosis and prediction method based on COSMIC data as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program or instruction stored therein, characterized in that: The computer program or instruction is programmed or configured to execute the boundary layer height diagnosis and prediction method based on COSMIC data as claimed in any one of claims 1 to 7 through a processor.

10. A computer program product comprising a computer program or instructions, characterized in that The computer program or instruction is programmed or configured to execute the boundary layer height diagnosis and prediction method based on COSMIC data as claimed in any one of claims 1 to 7 through a processor.

Images