A PM2.5 index based on weather circulation and stability index 2.5 Forecasting methods
By combining a PM2.5 forecasting method based on weather circulation and stability index with atmospheric science theory, a historical dataset of PM2.5 concentrations was established, which solved the problems of high computing power requirements of traditional methods and lack of interpretability of machine learning, and achieved efficient and interpretable PM2.5 forecasting.
Patent Information
- Application Number
- CN202411901683.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-20
- Publication Date
- 2025-09-30
- Estimated Expiration
- 2044-12-20
AI Technical Summary
Traditional numerical forecasting methods require high computing power and limited spatiotemporal resolution in PM2.5 pollution forecasting. Machine learning methods lack physical meaning and are difficult to apply in business applications.
The PM2.5 forecasting method based on weather circulation and stability index obtains monitoring data of the target area and meteorological observation data for future periods, calculates atmospheric circulation characteristics and stability characteristics, establishes a historical data set of PM2.5 concentration, and uses a lookup table method to predict PM2.5 concentration in future periods.
It shortens the forecast time, saves costs, and improves the interpretability of the forecast results. It combines atmospheric science theory to retain weather circulation and atmospheric stability diagnostic variables.
Smart Images
Figure CN119717073B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of air pollution prevention and control technology, in particular to a PM2.5 based on weather circulation and stability index. 2.5 Forecasting method. Background Art
[0002] PM 2.5 Pollution forecasting has always been a focus of urban air pollution prevention and control. Traditional numerical forecasting methods require high computing power, and the temporal and spatial resolution of forecast results is limited. In recent years, machine learning methods have been increasingly used in the field of atmospheric science. Although the forecast results of machine learning have improved to a certain extent compared with traditional numerical forecasts, their results lack physical meaning and are difficult to apply in business operations. Therefore, how to use the theoretical basis of atmospheric science as a starting point, combine long-term meteorological and pollutant concentration data sets, and achieve more accurate and simple PM2.5 predictions through modeling? 2.5 Pollution forecasting deserves further study. Summary of the Invention
[0003] The purpose of this application is to provide a PM based on weather circulation and stability index. 2.5 The forecasting method can shorten the forecasting time and improve the interpretability of the forecast results.
[0004] To achieve the above objectives, this application provides the following solutions:
[0005] In the first aspect, the present application provides a PM2.5 based on weather circulation and stability index. 2.5 Forecast method, the PM based on weather circulation and stability index 2.5 Forecasting methods include:
[0006] Obtain monitoring data of the target area and meteorological observation data for future periods; the monitoring data of the target area include: gridded sea level pressure data of the target area, ground wind speed data of the target area meteorological station, vertical temperature and dew point temperature data of the target area, and PM at each monitoring point in the target area. 2.5 Concentration time series; the gridded sea level pressure data of the target area is a three-dimensional matrix containing longitude, latitude and time dimensions; the ground wind speed data of the meteorological stations in the target area are the hourly ground wind speed observation data of each meteorological station; the meteorological observation data for the future period include: gridded sea level pressure data for the future period, ground wind speed data of the meteorological stations for the future period, and vertical temperature and dew point temperature data for the future period.
[0007] Keeping the time dimension unchanged, the average sea level pressure of the target area is subtracted from the gridded sea level pressure data of the target area to obtain the sea level pressure anomaly field of the target area.
[0008] The sea level pressure anomaly field of the target area is clustered to obtain a clustering result having the same length as the time dimension; the clustering result is a weather type.
[0009] The recirculation factor (RF) value of each meteorological station is calculated based on the ground wind speed data of the meteorological station in the target area.
[0010] The atmospheric circulation characteristics of the target area are determined based on the circulation index values of each meteorological station; the atmospheric circulation characteristics are circulation types.
[0011] The atmospheric stability index of the target area is calculated according to the vertical temperature and dew point temperature data of the target area.
[0012] The atmospheric stability characteristics of the target area are determined according to the atmospheric stability index of the target area; the atmospheric stability characteristics are K index intervals.
[0013] According to the PM of each monitoring point in the target area 2.5 The concentration time series is constructed by matching the weather type, circulation type and K index interval of the corresponding time. 2.5 Concentration history dataset.
[0014] Based on the meteorological observation data for the future period, the weather type, circulation type and K index range for the future period are calculated.
[0015] According to the weather type, circulation type and K index range of the future period, find the PM 2.5 Concentration historical data set, get PM 2.5 Concentration prediction results.
[0016] Optionally, keeping the time dimension unchanged, subtracting the average sea level pressure of the target area from the gridded sea level pressure data of the target area to obtain the sea level pressure anomaly field of the target area, specifically including:
[0017] The ERA5 sea level pressure data is divided by time to obtain hourly sea level pressure data.
[0018] The hourly sea level pressure anomaly data of the target area is obtained by subtracting the average value of all elements from each element in the hourly sea level pressure data of the target area.
[0019] The sea level pressure anomaly data at all times are grouped together every 24 hours and the average value is calculated to obtain the sea level pressure anomaly field of the target area.
[0020] Optionally, clustering the sea level pressure anomaly field of the target area to obtain a clustering result having the same length as the time dimension specifically includes:
[0021] The sea level pressure anomaly field of the target area is input into a SOM (Self-Organizing Map) neural network to obtain an identification matrix. The SOM neural network is composed of a set of two-dimensional matrices called nodes or neurons, each of which is connected to all neurons in the input layer and has a weight vector with the same dimension as the input data. The identification matrix includes the clustering results at each moment.
[0022] Optionally, the calculation formula of the circulation index value is:
[0023]
[0024] Among them, RF is the circulation index value, u j is the longitudinal wind speed at the jth moment, v j is the zonal wind speed at the jth moment, and τ is the total time.
[0025] Optionally, determining the atmospheric circulation characteristics of the target area according to the circulation index values of each meteorological station specifically includes:
[0026] For each time dimension, the mean and standard deviation of the circulation index values of each meteorological station are calculated.
[0027] When the average value of the circulation index is less than or equal to 0.2 and the standard deviation of the circulation index is less than or equal to 0.1, the atmospheric circulation characteristics are persistent circulation.
[0028] When the average value of the circulation index is greater than or equal to 0.3 and the standard deviation of the circulation index is less than or equal to 0.15, the atmospheric circulation characteristics are rotational circulation.
[0029] When the standard deviation of the circulation index value is greater than 0.15, the atmospheric circulation characteristics are discrete circulation.
[0030] When the average value of the circulation index is greater than 0.2 and less than 0.3, and the standard deviation of the circulation index is greater than 0.1 and less than 0.15, the atmospheric circulation characteristics are mixed circulation.
[0031] Optionally, the calculation formula of the atmospheric stability index is:
[0032] K=(T 1000 -T 850 )+T d1000 -(T 925 -T d925 );
[0033] Among them, K is the atmospheric stability index, T 1000 is the temperature of the 1000 geopotential layer, T 850 is the temperature of the 850° geopotential layer, T d1000is the dew point temperature at 1000 geopotential level, T 925 is the dew point temperature of the 925 geopotential layer, T d925 It is the dew point temperature at the 925 geopotential layer.
[0034] Optionally, determining the atmospheric stability characteristics of the target area according to the atmospheric stability index of the target area specifically includes:
[0035] For each time dimension, the average value of the K index of each meteorological station is calculated.
[0036] When the average value of the K index is greater than -40 and less than or equal to -30, the K index interval is 1.
[0037] When the average value of the K index is greater than -30 and less than or equal to -20, the K index interval is 2.
[0038] When the average value of the K index is greater than -20 and less than or equal to -10, the K index interval is 3.
[0039] When the average value of the K index is greater than -10 and less than or equal to 0, the K index interval is 4.
[0040] When the average value of the K index is greater than 0 and less than or equal to 10, the K index interval is 5.
[0041] When the average value of the K index is greater than 10 and less than or equal to 20, the K index interval is 6.
[0042] When the average value of the K index is greater than 20 and less than or equal to 30, the K index interval is 7.
[0043] When the average value of the K index is greater than 30 and less than or equal to 40, the K index interval is 8.
[0044] When the average value of the K index is greater than 40 or less than or equal to -40, the K index interval is 9.
[0045] Optionally, according to the PM of each monitoring point in the target area 2.5 The concentration time series is constructed by matching the weather type, circulation type and K index interval of the corresponding time. 2.5 Concentration history data set, including:
[0046] According to weather type, circulation type and K index range, PM2.5 at each monitoring point is calculated. 2.5 PM concentration time series corresponding to 2.5 The concentrations are stored in the matrix pm.
[0047] Let the matrix pm be PM 2.5 Concentration history dataset.
[0048] Optionally, based on the meteorological observation data for the future period, calculating the weather type, circulation type and K index range for the future period specifically includes:
[0049] Keeping the time dimension unchanged, the gridded sea level pressure data of the future period is subtracted from the average sea level pressure of the future period to obtain the sea level pressure anomaly field of the future period.
[0050] Based on the sea level pressure anomaly field in the future period, the distance to each weather type is calculated separately.
[0051] The weather type with the smallest distance is used as the weather type for the future period.
[0052] Calculate the circulation index value for the future period based on the ground wind speed data of the meteorological station for the future period.
[0053] The circulation type of the future period is determined according to the circulation index value of the future period.
[0054] The atmospheric stability index of the future period is calculated according to the vertical temperature and dew point temperature data of the future period.
[0055] A K index range for the future period is determined according to the atmospheric stability index for the future period.
[0056] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0057] This application provides a PM based on weather circulation and stability index. 2.5 The forecasting method comprises the following steps: first, obtaining monitoring data of the target area and meteorological observation data for the future period; keeping the time dimension unchanged, subtracting the average sea level pressure of the target area from the gridded sea level pressure data of the target area to obtain the sea level pressure anomaly field of the target area; secondly, clustering the sea level pressure anomaly field of the target area to obtain a clustering result with the same length as the time dimension; the clustering result is a weather classification; according to the ground wind speed data of the meteorological station in the target area, the circulation index value of each meteorological station is calculated; according to the circulation index value of each meteorological station, the atmospheric circulation characteristics of the target area are determined; the atmospheric circulation characteristics are circulation classification; according to the vertical temperature and dew point temperature data of the target area, the atmospheric stability index of the target area is calculated; according to the atmospheric stability index of the target area, the atmospheric stability characteristics of the target area are determined; the atmospheric stability characteristics are K index intervals; then, according to the PM values of each monitoring point in the target area, the atmospheric stability index of the target area is calculated. 2.5 The concentration time series is constructed by matching the weather type, circulation type and K index interval of the corresponding time. 2.5Concentration historical data set; finally, based on the meteorological observation data of the future period, calculate the weather type, circulation type and K index range of the future period; based on the weather type, circulation type and K index range of the future period, find PM 2.5 Concentration historical data set, get PM 2.5 Compared with numerical prediction, this application establishes PM 2.5 Concentration historical data sets, using the “lookup table method” based on historical real data sets to predict PM2.5 concentrations in future periods 2.5 concentration, shortening the forecast time and greatly saving time and labor costs; compared with the machine learning prediction of PM based on historical observation data 2.5 Compared with the concentration method, this application models the PM from the theoretical basis of atmospheric science. 2.5 In addition to the concentration prediction results, diagnostic variables such as weather circulation and atmospheric stability are also retained, which improves the interpretability of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0059] Figure 1 In one embodiment of the present application, a PM2.5 index based on weather circulation and stability index is 2.5 Diagram of the application environment of the forecast method.
[0060] Figure 2 A PM2.5 based on weather circulation and stability index is provided in one embodiment of the present application. 2.5 Schematic diagram of the overall process of the forecasting method.
[0061] Figure 3 A PM2.5 based on weather circulation and stability index is provided in one embodiment of the present application. 2.5 Flowchart of the forecasting method. DETAILED DESCRIPTION
[0062] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.
[0063] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0064] The PM based on weather circulation and stability index provided in the embodiment of the present application 2.5 The forecasting method can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the acquired monitoring data of the target area and the meteorological observation data for the future period to the server 104. The monitoring data of the target area include: gridded sea level pressure data of the target area, ground wind speed data of the meteorological station in the target area, vertical temperature and dew point temperature data of the target area, and PM of each monitoring point in the target area. 2.5 concentration time series; the gridded sea level pressure data of the target area is a three-dimensional matrix including longitude, latitude and time dimensions; the surface wind speed data of the meteorological station in the target area is the hourly surface wind speed observation data of each meteorological station; the meteorological observation data for the future period includes: gridded sea level pressure data for the future period, surface wind speed data of the meteorological station for the future period, and vertical temperature and dew point temperature data for the future period; after the server 104 receives the monitoring data of the target area and the meteorological observation data for the future period, for the monitoring data and the meteorological observation data for the future period, the server 104 keeps the time dimension unchanged and subtracts the gridded sea level pressure data of the target area from the surface wind speed data of the target area. The sea level pressure anomaly field of the target area is obtained by clustering the sea level pressure anomaly field of the target area to obtain a clustering result of the same length as the time dimension; the clustering result is a weather type; the circulation index value of each meteorological station is calculated according to the ground wind speed data of the meteorological station in the target area; the atmospheric circulation characteristics of the target area are determined according to the circulation index value of each meteorological station; the atmospheric circulation characteristics are circulation types; the atmospheric stability index of the target area is calculated according to the vertical temperature and dew point temperature data of the target area; the atmospheric stability characteristics of the target area are determined according to the atmospheric stability index of the target area; the atmospheric stability characteristics of the target area are K index intervals; the PM values of each monitoring point in the target area are calculated according to the atmospheric stability index of the target area. 2.5 The concentration time series is constructed by matching the weather type, circulation type and K index interval of the corresponding time. 2.5 Concentration historical data set; based on the meteorological observation data of the future period, calculate the weather type, circulation type and K index range of the future period; based on the weather type, circulation type and K index range of the future period, find the PM 2.5Concentration historical data set, get PM 2.5 The server 104 can obtain the PM 2.5 The concentration prediction result is fed back to the terminal 102. In addition, in some embodiments, PM2.5 concentration is predicted based on the weather circulation and stability index. 2.5 The forecast method can also be implemented by the server 104 or the terminal 102 alone. For example, the terminal 102 can directly perform PM prediction based on the monitoring data of the target area and the meteorological observation data of the future period. 2.5 Forecast, the server 104 can also obtain the monitoring data of the target area and the meteorological observation data of the future period from the data storage system, and perform PM on the monitoring data of the target area and the meteorological observation data of the future period. 2.5 forecast.
[0065] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, and tablet computers. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or a cloud server.
[0066] In an exemplary embodiment, Figure 2 and Figure 3 As shown, a PM2.5 index based on weather circulation and stability index is provided. 2.5 The forecast method is executed by a computer device, specifically a terminal or a server, etc., and can be executed alone, or jointly by a terminal and a server. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate the method, which includes the following steps S1 to S10.
[0067] S1: Acquire the monitoring data of the target area and the meteorological observation data for the future period; the monitoring data of the target area include: gridded sea level pressure data of the target area, ground wind speed data of the meteorological station in the target area, vertical temperature and dew point temperature data of the target area, and PM at each monitoring point in the target area. 2.5 Concentration time series; the gridded sea level pressure data of the target area is a three-dimensional matrix containing longitude, latitude and time dimensions; the ground wind speed data of the meteorological stations in the target area are the hourly ground wind speed observation data of each meteorological station; the meteorological observation data for the future period include: gridded sea level pressure data for the future period, ground wind speed data of the meteorological stations for the future period, and vertical temperature and dew point temperature data for the future period.
[0068] S2: Keeping the time dimension unchanged, subtract the average sea level pressure of the target area from the gridded sea level pressure data of the target area to obtain the sea level pressure anomaly field of the target area.
[0069] S3: Clustering the sea level pressure anomaly field of the target area to obtain a clustering result with the same length as the time dimension; the clustering result is a weather type.
[0070] S4: Calculate the circulation index value of each meteorological station according to the ground wind speed data of the meteorological station in the target area.
[0071] S5: Determine the atmospheric circulation characteristics of the target area according to the circulation index values of each meteorological station; the atmospheric circulation characteristics are circulation types.
[0072] S6: Calculate the atmospheric stability index of the target area according to the vertical temperature and dew point temperature data of the target area.
[0073] S7: Determine an atmospheric stability characteristic of the target area according to the atmospheric stability index of the target area; the atmospheric stability characteristic is a K index interval.
[0074] S8: Based on the PM values of each monitoring point in the target area 2.5 The concentration time series is constructed by matching the weather type, circulation type and K index interval of the corresponding time. 2.5 Concentration history dataset.
[0075] S9: Calculate the weather type, circulation type and K index range of the future period based on the meteorological observation data of the future period.
[0076] S10: Find the PM according to the weather type, circulation type and K index range of the future period 2.5 Concentration historical data set, get PM 2.5 Concentration prediction results.
[0077] Implement the above steps S1 to S10. Compared with numerical forecast, this application establishes PM 2.5 Concentration historical data sets, using the “lookup table method” based on historical real data sets to predict PM2.5 concentrations in future periods 2.5 concentration, shortening the forecast time and greatly saving time and labor costs; compared with the machine learning prediction of PM based on historical observation data 2.5 Compared with the concentration method, this application models the PM from the theoretical basis of atmospheric science. 2.5 In addition to the concentration prediction results, diagnostic variables such as weather circulation and atmospheric stability are also retained, which improves the interpretability of the prediction results.
[0078] The following description is based on the PM2.5 and PM2.5 stability indices. 2.5 The forecast method uses ERA5 reanalysis data, observation data of each meteorological station in the target area, and historical PM values at each monitoring point. 2.5Data to predict future PM 2.5 Example of concentration.
[0079] In an exemplary embodiment, step S2 specifically includes:
[0080] S21: Divide the ERA5 sea level pressure data by time to obtain hourly sea level pressure data.
[0081] S22: Subtract the average value of all elements from each element in the hourly sea level pressure data of the target area to obtain the hourly sea level pressure anomaly data.
[0082] S23 calculates the average value of the sea level pressure anomaly data at all times every 24 hours to obtain the sea level pressure anomaly field of the target area.
[0083] Specifically, first divide the ERA5 sea level pressure data by time, and calculate the sea level pressure data P of the target area (M×N longitude and latitude grid) at the i-th hour. i Each element in (M×N matrix) is subtracted from the mean value p of all its elements i , get the sea level pressure anomaly data D at the i-th hour i , calculate the average value of the sea level pressure anomaly data at all times every 24 hours as a group, and reorganize it into an M×N×X sea level pressure anomaly field D1…D j …D X .
[0084] In an exemplary embodiment, step S3 specifically includes:
[0085] The sea level pressure anomaly field of the target area is input into a SOM neural network to obtain an identification matrix; the SOM neural network is composed of a set of two-dimensional matrices called nodes or neurons, each neuron is connected to all neurons in the input layer and has a weight vector with the same dimension as the input data; the identification matrix includes the clustering results at each moment.
[0086] Specifically, a SOM neural network is typically composed of a set of two-dimensional matrices called nodes or neurons. Each neuron is connected to all neurons in the input layer and has a weight vector with the same dimension as the input data. The training principle is to continuously adjust these weight vectors so that they can map and characterize the characteristics of the input data. Using an unsupervised training method based on competitive learning, the SOM neural network can map high-dimensional input data onto a common two-dimensional topological structure while maintaining the topological relationships of the original data. This makes it ideal for clustering and visualizing high-dimensional data. The sea level pressure anomaly field is a three-dimensional matrix of size M×N×X.
[0087] In this example, a 3×3 SOM neural network was used. The input was an M×N×X sea level pressure anomaly field, and the output was an X×1 identification matrix C, representing the clustering results for each moment, for a total of nine categories. The sea level pressure anomaly fields belonging to the same category were averaged to obtain nine M×N matrices, NODE1, …, NODE9, which are called weather types. For example, at time e, we have:
[0088] NODE1: C(e)=1; NODE2: C(e)=2; NODE3: C(e)=3;
[0089] NODE4: C(e)=4; NODE5: C(e)=5; NODE6: C(e)=6;
[0090] NODE7: C(e)=7; NODE8: C(e)=8; NODE9: C(e)=9.
[0091] This embodiment classifies atmospheric circulations of different scales according to the local circulation index at different meteorological stations. The specific calculation formula of the circulation index value is as follows based on the ratio of air vector displacement to scalar displacement:
[0092]
[0093] Among them, RF is the circulation index value, u j is the longitudinal wind speed at the jth moment, v j is the zonal wind speed at the jth moment, and τ is the total time.
[0094] In an exemplary embodiment, step S5 specifically includes:
[0095] S51: For each time dimension, calculate the mean and standard deviation of the circulation index values of each meteorological station.
[0096] S52: When the average value of the circulation index is less than or equal to 0.2 and the standard deviation of the circulation index is less than or equal to 0.1, the atmospheric circulation characteristics are persistent circulation.
[0097] S53: When the average value of the circulation index is greater than or equal to 0.3 and the standard deviation of the circulation index is less than or equal to 0.15, the atmospheric circulation is characterized by a rotational circulation.
[0098] S54: When the standard deviation of the circulation index value is greater than 0.15, the atmospheric circulation characteristics are discrete circulation.
[0099] S55: When the average value of the circulation index is greater than 0.2 and less than 0.3, and the standard deviation of the circulation index is greater than 0.1 and less than 0.15, the atmospheric circulation characteristics are mixed circulation.
[0100] In this embodiment, τ is set to 24 hours. The calculation result RF represents the recirculation degree of the wind field at the meteorological station, and its value range is [0, 1]. A low RF value represents a lower recirculation degree, that is, a more ventilated condition; conversely, it represents an unfavorable ventilation condition. The wind speed data of I meteorological stations in the target area are calculated to obtain I time series RF1,..., RF of X×1. I For each time dimension, calculate the average value RF_avg and the standard deviation RF_std of RF for I meteorological stations, and classify them according to the following criteria:
[0101] RF_avg ≤ 0.2, RF_std ≤ 0.1: Continuous circulation;
[0102] RF_avg ≥ 0.3, RF_std ≤ 0.15: Rotating circulation;
[0103] RF_std > 0.15: Discrete circulation;
[0104] 0.2 < RF_avg < 0.3, 0.1 < RF_std < 0.15: Mixed circulation.
[0105] Thus, an identification matrix RF_type of X×1 is obtained, representing the classification result at each moment, which is called circulation classification, with a total of 4 categories.
[0106] The K-index is an instability index that considers the water vapor conditions in the air layer. As an empirical index, it reflects both the atmospheric stratification stability and the water vapor conditions in the middle and lower layers, with the unit of °C. In this embodiment, the expression of the atmospheric stability index is as follows:
[0107] K = (T 1000 - T 850 ) + T d1000 - (T 925 - T d925 ) (2);
[0108] where K is the atmospheric stability index, T 1000 is the temperature at the 1000 geopotential height level, T 850 is the temperature at the 850 geopotential height level, T d1000 is the dew point temperature at the 1000 geopotential height level, T 925 is the dew point temperature at the 925 geopotential height level, T d925 is the dew point temperature at the 925 geopotential height level.
[0109] It should be noted that the unit of the corresponding geopotential height level is hPa.
[0110] In an exemplary embodiment, in step S7, it specifically includes:
[0111] S71: For each time dimension, calculate the average value of the K index of each meteorological station.
[0112] S72: When the average value of the K index is greater than -40 and less than or equal to -30, the K index interval is 1.
[0113] S73: When the average value of the K index is greater than -30 and less than or equal to -20, the K index interval is 2.
[0114] S74: When the average value of the K index is greater than -20 and less than or equal to -10, the K index interval is 3.
[0115] S75: When the average value of the K index is greater than -10 and less than or equal to 0, the K index interval is 4.
[0116] S76: When the average value of the K index is greater than 0 and less than or equal to 10, the K index interval is 5.
[0117] S77: When the average value of the K index is greater than 10 and less than or equal to 20, the K index interval is 6.
[0118] S78: When the average value of the K index is greater than 20 and less than or equal to 30, the K index interval is 7.
[0119] S79: When the average value of the K index is greater than 30 and less than or equal to 40, the K index interval is 8.
[0120] S710: When the average value of the K index is greater than 40 or less than or equal to -40, the K index range is 9.
[0121] Specifically, the K index of I meteorological stations in the target area is calculated according to Formula 2, and I X×1 time series K1,…,K I For each time dimension, the K index average value K_avg (X×1 matrix) of I meteorological stations is calculated. This embodiment classifies according to the following criteria:
[0122] -40 <K_avg≤-30,K_type=1;
[0123] -30 <K_avg≤-20,K_type=2;
[0124] -20 <K_avg≤-10,K_type=3;
[0125] -10 <K_avg≤0,K_type=4;
[0126] 0 <K_avg≤10,K_type=5;
[0127] 10 <K_avg≤20,K_type=6;
[0128] 20 <K_avg≤30,K_type=7;
[0129] 30 <K_avg≤40,K_type=8;
[0130] K_avg>40 or K_avg≤-40, K_type=9.
[0131] Thus, we get the X×1 identification matrix K_type, which represents the range of K index values at each moment, called K index interval, with a total of 9 categories.
[0132] In an exemplary embodiment, step S8 specifically includes:
[0133] S81: According to weather type, circulation type and K index range, PM2.5 at each monitoring point is calculated. 2.5 PM concentration time series corresponding to 2.5 The concentrations are stored in the matrix pm.
[0134] S82: Take the matrix pm as PM 2.5 Concentration history dataset.
[0135] Specifically, the PM2.5 concentration time series (X×R matrix) of the R monitoring points in the target area are averaged into an X×1 PM2.5 concentration time series CONC. According to the classification results of the identification matrix C, RF_type, and K_type, the PM2.5 concentration at the time corresponding to CONC is stored in a 9×4×9×X matrix pm. The matrix pm is called the PM2.5 concentration historical dataset.
[0136] In an exemplary embodiment, step S8 specifically includes:
[0137] S81: Keeping the time dimension unchanged, subtracting the average sea level pressure of the future period from the gridded sea level pressure data of the future period to obtain the sea level pressure anomaly field of the future period.
[0138] S82: Calculate the distance to each weather type based on the sea level pressure anomaly field in the future period.
[0139] S83: The weather type with the smallest distance is used as the weather type for the future period.
[0140] S84: Calculate the circulation index value for the future period based on the ground wind speed data of the meteorological station for the future period.
[0141] S85: Determine the circulation type of the future period according to the circulation index value of the future period.
[0142] S86: Calculating the atmospheric stability index for the future period based on the vertical temperature and dew point temperature data for the future period.
[0143] S87: Determine a K index range for the future period according to the atmospheric stability index for the future period.
[0144] Specifically, for the future sea level pressure anomaly fields SLP1,…,SLP t (M×N×t matrix, t is in days), calculate the sea level pressure anomaly field SLP on the kth day k Distance d from weather type NODE1, ..., NODE9 1k ,…,d 9k , take the NODE type corresponding to the smallest distance as the weather type of day k. For example, for day k, min{d 1k ,…,d 9k}=d 3k , the weather type of day k is NODE3, which is stored in the matrix node and recorded as node(k) = 3. Thus, the weather type node (t×1 matrix) of the target area in the future period is obtained.
[0145] According to the above method of extracting the atmospheric circulation characteristics of the target area using the circulation index, the identification matrix rf_type (t×1 matrix) of the future time period is calculated.
[0146] According to the above method of using the K index to extract the atmospheric stability characteristics of the target area, the identification matrix k_type (t×1 matrix) of the future period is calculated.
[0147] In step S10, it specifically includes:
[0148] According to the identification matrix node, rf_type, k_type of the future time period, the corresponding search is performed in the established PM2.5 concentration historical dataset pm. For example, the weather type of day k is node(k), the circulation type is rf_type(k), and the K index interval is k_type(k). Then the PM2.5 concentration sequence corresponding to node(k), rf_type(k), and k_type(k) in the PM2.5 concentration historical dataset pm matrix is Y k , take Y k The median y k is the predicted PM2.5 concentration of the target area on the kth day. Thus, the predicted PM2.5 concentration results y1,…,y for the target area in the future period are obtained. t Returns the user.
[0149] This application also provides an application scenario, which uses the PM based on weather circulation and stability index.2.5 Forecast method. Specifically: The PM2.5 prediction method based on weather circulation and stability index provided in this embodiment 2.5 Forecasting methods can be applied to PM 2.5 In the forecast scenario. PM 2.5 Forecast scenarios include: data acquisition, PM 2.5 The concentration historical data set establishment and prediction stages: First, obtain the monitoring data of the target area and the meteorological observation data for the future period; the monitoring data of the target area include: gridded sea level pressure data of the target area, ground wind speed data of the target area meteorological station, vertical temperature and dew point temperature data of the target area, and PM2.5 at each monitoring point in the target area. 2.5 Concentration time series; the gridded sea level pressure data of the target area is a three-dimensional matrix containing longitude, latitude and time dimensions; the surface wind speed data of the meteorological station in the target area is the hourly surface wind speed observation data of each meteorological station; the meteorological observation data for the future period include: gridded sea level pressure data for the future period, surface wind speed data of the meteorological station for the future period, and vertical temperature and dew point temperature data for the future period; secondly, keeping the time dimension unchanged, the gridded sea level pressure data of the target area is subtracted from the average sea level pressure of the target area to obtain the sea level pressure anomaly field of the target area; the sea level of the target area is calculated by The surface pressure anomaly field is clustered to obtain a clustering result of the same length as the time dimension; the clustering result is a weather type; based on the surface wind speed data of the meteorological station in the target area, the circulation index value of each meteorological station is calculated; based on the circulation index value of each meteorological station, the atmospheric circulation characteristics of the target area are determined; the atmospheric circulation characteristics are circulation types; based on the vertical temperature and dew point temperature data of the target area, the atmospheric stability index of the target area is calculated; based on the atmospheric stability index of the target area, the atmospheric stability characteristics of the target area are determined; the atmospheric stability characteristics are K index intervals; based on the PM values of each monitoring point in the target area, the atmospheric stability characteristics of the target area are determined; the atmospheric stability characteristics of the target area are determined; the atmospheric stability characteristics are K index intervals 2.5 The concentration time series is constructed by matching the weather type, circulation type and K index interval of the corresponding time. 2.5 Then, according to the meteorological observation data of the future period, the weather type, circulation type and K index interval of the future period are calculated; according to the weather type, circulation type and K index interval of the future period, the PM 2.5 Concentration historical data sets can be used to obtain PM 2.5 Concentration prediction results.
[0150] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0151] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A PM2.5 index based on weather circulation and stability index 2.5 The forecasting method is characterized in that The PM based on weather circulation and stability index 2.5 Forecasting methods include: Obtain monitoring data of the target area and meteorological observation data for future periods; the monitoring data of the target area include: gridded sea level pressure data of the target area, ground wind speed data of the target area meteorological station, vertical temperature and dew point temperature data of the target area, and PM at each monitoring point in the target area. 2.5 Concentration time series; the gridded sea level pressure data of the target area is a three-dimensional matrix containing longitude, latitude and time dimensions; the surface wind speed data of the meteorological stations in the target area is the hourly surface wind speed observation data of each meteorological station; the meteorological observation data for the future period includes: gridded sea level pressure data for the future period, surface wind speed data of the meteorological stations for the future period, and vertical temperature and dew point temperature data for the future period; Keeping the time dimension unchanged, subtracting the average sea level pressure of the target area from the gridded sea level pressure data of the target area to obtain the sea level pressure anomaly field of the target area; Clustering the sea level pressure anomaly field of the target area to obtain a clustering result having the same length as the time dimension; the clustering result is a weather type; Calculating the circulation index value of each meteorological station based on the ground wind speed data of the target area meteorological station; Determining atmospheric circulation characteristics of a target area according to the circulation index values of each meteorological station; the atmospheric circulation characteristics are circulation types; Calculating the atmospheric stability index of the target area based on the vertical temperature and dew point temperature data of the target area; determining an atmospheric stability characteristic of the target area according to the atmospheric stability index of the target area; wherein the atmospheric stability characteristic is a K index interval; According to the PM of each monitoring point in the target area 2.5 The concentration time series is constructed by matching the weather type, circulation type and K index interval of the corresponding time. 2.5 Concentration history dataset; Calculate the weather type, circulation type and K index range for the future period based on the meteorological observation data for the future period; According to the weather type, circulation type and K index range of the future period, find the PM 2.5 Concentration historical data set, get PM 2.5 Concentration prediction results.
2. PM based on weather circulation and stability index according to claim 1 2.5 The forecasting method is characterized in that Keeping the time dimension unchanged, subtract the average sea level pressure of the target area from the gridded sea level pressure data of the target area to obtain the sea level pressure anomaly field of the target area, specifically including: Divide the ERA5 sea level pressure data by time to obtain hourly sea level pressure data; Subtract the average value of all elements from each element in the hourly sea level pressure data of the target area to obtain the hourly sea level pressure anomaly data; The sea level pressure anomaly data at all times are grouped together every 24 hours and the average value is calculated to obtain the sea level pressure anomaly field of the target area.
3. PM based on weather circulation and stability index according to claim 1 2.5 The forecasting method is characterized in that Cluster the sea level pressure anomaly field of the target area to obtain clustering results with the same length as the time dimension, including: The sea level pressure anomaly field of the target area is input into a SOM neural network to obtain an identification matrix; the SOM neural network is composed of a set of two-dimensional matrices called nodes or neurons, each neuron is connected to all neurons in the input layer and has a weight vector with the same dimension as the input data; the identification matrix includes the clustering results at each moment.
4. PM based on weather circulation and stability index according to claim 1 2.5 The forecasting method is characterized in that The calculation formula of the circulation index value is: Among them, RF is the circulation index value, u j is the longitudinal wind speed at the jth moment, v j is the zonal wind speed at the jth moment, and τ is the total time.
5. The PM2.5 based on weather circulation and stability index according to claim 1 2.5 The forecasting method is characterized in that Determine the atmospheric circulation characteristics of the target area based on the circulation index values of each meteorological station, specifically including: For each time dimension, the mean and standard deviation of the circulation index values at each meteorological station are calculated; When the average value of the circulation index is less than or equal to 0.2, and the standard deviation of the circulation index is less than or equal to 0.1, the atmospheric circulation characteristics are persistent circulation; When the average value of the circulation index is greater than or equal to 0.3 and the standard deviation of the circulation index is less than or equal to 0.15, the atmospheric circulation is characterized by a rotational circulation; When the standard deviation of the circulation index value is greater than 0.15, the atmospheric circulation is characterized by a discrete circulation; When the average value of the circulation index is greater than 0.2 and less than 0.3, and the standard deviation of the circulation index is greater than 0.1 and less than 0.15, the atmospheric circulation characteristics are mixed circulation.
6. The PM2.5 based on weather circulation and stability index according to claim 1 2.5 The forecasting method is characterized in that The calculation formula of the atmospheric stability index is: K=(T 1000 -T 850 )+T d1000 -(T 925 -T d925 ); Among them, K is the atmospheric stability index, T 1000 is the temperature of the 1000 geopotential layer, T 850 is the temperature of the 850° geopotential layer, T d1000 is the dew point temperature at 1000 geopotential level, T 925 is the dew point temperature of the 925 geopotential layer, T d925 It is the dew point temperature at the 925 geopotential layer.
7. The PM2.5 based on weather circulation and stability index according to claim 1 2.5 The forecasting method is characterized in that Determining the atmospheric stability characteristics of the target area according to the atmospheric stability index of the target area includes: For each time dimension, the average value of the K index of each meteorological station is calculated; When the average value of the K index is greater than -40 and less than or equal to -30, the K index interval is 1; When the average value of the K index is greater than -30 and less than or equal to -20, the K index interval is 2; When the average value of the K index is greater than -20 and less than or equal to -10, the K index interval is 3; When the average value of the K index is greater than -10 and less than or equal to 0, the K index interval is 4; When the average value of the K index is greater than 0 and less than or equal to 10, the K index interval is 5; When the average value of the K index is greater than 10 and less than or equal to 20, the K index interval is 6; When the average value of the K index is greater than 20 and less than or equal to 30, the K index interval is 7; When the average value of the K index is greater than 30 and less than or equal to 40, the K index interval is 8; When the average value of the K index is greater than 40 or less than or equal to -40, the K index interval is 9.
8. The PM2.5 based on weather circulation and stability index according to claim 1 2.5 The forecasting method is characterized in that According to the PM of each monitoring point in the target area 2.5 The concentration time series is constructed by matching the weather type, circulation type and K index interval of the corresponding time. 2.5 Concentration history data set, including: According to weather type, circulation type and K index range, PM2.5 at each monitoring point is calculated. 2.5 PM concentration time series corresponding to 2.5 The concentration is stored in the matrix pm; Let the matrix pm be PM 2.5 Concentration history dataset.
9. The PM2.5 based on weather circulation and stability index according to claim 1 2.5 The forecasting method is characterized in that Based on the meteorological observation data for the future period, the weather type, circulation type and K index range for the future period are calculated, specifically including: Keeping the time dimension unchanged, subtracting the average sea level pressure of the future period from the gridded sea level pressure data of the future period to obtain the sea level pressure anomaly field of the future period; Calculate the distance to each weather type based on the sea level pressure anomaly field in the future period; The weather type with the smallest distance is used as the weather type for the future period; Calculating a circulation index value for the future period based on ground wind speed data from a meteorological station for the future period; Determining the circulation type of the future period according to the circulation index value of the future period; Calculating an atmospheric stability index for the future period based on vertical temperature and dew point temperature data for the future period; A K index range for the future period is determined according to the atmospheric stability index for the future period.