Seasonal climate statistical prediction method, system, device and storage medium

Abstract

This application discloses a seasonal climate statistical prediction method to address existing climate statistical prediction methods, such as the lack of targeted optimization of characteristic factors and models, simple characteristic information and regression algorithms, and poor spatial distribution prediction techniques for climate anomalies. The method comprises: obtaining meteorological data within a specified time period; preprocessing the meteorological data to obtain forecast object data, and performing principal component decomposition on the forecast object data to obtain at least one forecast quantity; obtaining monthly data for specified variables at global grid points, performing time-dimensional averaging and difference preprocessing on the specified variable data to obtain multiple global-scale characteristic information fields; using the leave-one-out method to calculate the correlation coefficient between each forecast quantity and each global-scale characteristic information field, and establishing a forecast factor based on the location of the maximum correlation coefficient; and using a distributed gradient boosting algorithm to statistically model the forecast factor to obtain a climate prediction model, thereby predicting the climate of the forecast year.

Description

Technical Field

[0001] The present application relates to the field of weather forecasting, and in particular to a seasonal climate statistical forecasting method, system, device and storage medium. Background Art

[0002] At present, our short-term climate forecasting work is based on two aspects: one is to establish computerized climate models and data assimilation systems based on various observational data and climate system equations and use them for forecasting; the other is to establish forecast factors and statistical models based on empirical methods based on historical station observation data and atmospheric and ocean circulation data and make predictions for the future.

[0003] However, current statistical methods and climate model predictions generally have poor accuracy in predicting the spatial distribution of climate anomalies. Existing statistical methods, when developing forecast factors and training models, are often optimized for historical periods, rather than for the year being predicted. This limits their predictive power given the often significant interdecadal variability. Furthermore, compared to machine learning, existing empirical statistical methods are overly simplistic in their processing of forecast information sources and regression methods.

[0004] It can be seen from this that how to improve traditional empirical statistics and provide a feasible climate statistical prediction solution to improve climate prediction capabilities, especially the ability to predict the spatial distribution of climate anomalies, has become an urgent problem to be solved by technical personnel in this field. Summary of the Invention

[0005] The embodiment of the present application provides a seasonal climate statistical prediction method to solve the problems of existing climate statistical prediction methods, such as the lack of targeted feature factor and model optimization, simple feature information and regression algorithms, and poor spatial distribution prediction skills of climate anomalies.

[0006] The embodiment of the present application also provides a seasonal climate statistical prediction system to solve the problems of existing climate statistical prediction methods, such as the lack of targeted feature factor and model optimization, simple feature information and regression algorithms, and poor spatial distribution prediction skills of climate anomalies.

[0007] The embodiment of the present application also provides a seasonal climate statistical prediction device to solve the problems of existing climate statistical prediction methods, such as the lack of targeted feature factor and model optimization, simple feature information and regression algorithms, and poor spatial distribution prediction skills of climate anomalies.

[0008] The embodiment of the present application also provides a computer-readable storage medium to solve the problems of existing climate statistical prediction methods, such as the lack of targeted feature factor and model optimization, simple feature information and regression algorithms, and poor spatial distribution prediction skills of climate anomalies.

[0009] The embodiments of this application adopt the following technical solutions:

[0010] A seasonal climate statistical prediction method, comprising:

[0011] According to the determined year to be predicted, meteorological data within a specified time period is obtained, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; the meteorological data is preprocessed to obtain forecast object data, and the forecast object data is decomposed into principal components to obtain at least one forecast quantity; designated variable data of each month at a global grid point within the specified time period is obtained, and the designated variable data is preprocessed in terms of time dimension average and difference to obtain at least one global-scale characteristic information field; the correlation coefficient between each forecast quantity and each global-scale characteristic information field is calculated using the leave-one-out method, and a forecast factor is established according to the position of the maximum value of the correlation coefficient; the forecast factor is statistically modeled using a distributed gradient boosting algorithm to obtain a climate prediction model, and the climate of the year to be predicted is predicted based on the climate prediction model.

[0012] A seasonal climate statistical prediction system comprises: a meteorological data acquisition unit for acquiring meteorological data within a specified time period according to a determined year to be predicted, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; a preprocessing unit for preprocessing the meteorological data to obtain forecast object data, and performing principal component decomposition on the forecast object data to obtain at least one forecast quantity; a characteristic information field determination unit for acquiring designated variable data for each month at a global grid point within the specified time period, performing time-dimensional averaging and difference preprocessing on the designated variable data to obtain at least one global-scale characteristic information field; a prediction factor determination unit for calculating the correlation coefficient between each forecast quantity and each global-scale characteristic information field using a leave-one-out method, and establishing a prediction factor based on the position of the maximum value of the correlation coefficient; and a prediction unit for performing statistical modeling on the prediction factor using a distributed gradient boosting algorithm to obtain a climate prediction model, and predicting the climate of the year to be predicted based on the climate prediction model.

[0013] A seasonal climate statistical prediction device, comprising:

[0014] A processor; and a memory arranged to store computer-executable instructions, wherein the executable instructions, when executed, cause the processor to perform the following operations: according to the determined year to be predicted, obtain meteorological data within a specified time period, wherein the meteorological data include monthly average precipitation data or monthly average temperature data for each month within the specified time period; preprocess the meteorological data to obtain forecast object data, and perform principal component decomposition on the forecast object data to obtain at least one forecast quantity; obtain designated variable data for each month of the global grid within the specified time period, perform time-dimensional average and difference preprocessing on the designated variable data to obtain at least one global-scale characteristic information field; use the leave-one-out method to calculate the correlation coefficient between each forecast quantity and each global-scale characteristic information field, and establish a forecast factor according to the position of the maximum value of the correlation coefficient; use a distributed gradient boosting algorithm to perform statistical modeling on the forecast factor to obtain a climate prediction model, and predict the climate of the year to be predicted based on the climate prediction model.

[0015] A computer-readable storage medium stores one or more programs, which, when executed by an electronic device including multiple application programs, cause the electronic device to perform the following operations: obtain meteorological data within a specified time period according to a determined year to be predicted, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; preprocess the meteorological data to obtain forecast object data, and perform principal component decomposition on the forecast object data to obtain at least one forecast quantity; obtain designated variable data for each global grid point within the specified time period, perform time-dimensional averaging and difference preprocessing on the designated variable data to obtain at least one global-scale characteristic information field; use the leave-one-out method to calculate the correlation coefficient between each forecast quantity and each global-scale characteristic information field, and establish a forecast factor based on the position of the maximum value of the correlation coefficient; use a distributed gradient boosting algorithm to perform statistical modeling on the forecast factor to obtain a climate prediction model, and predict the climate of the year to be predicted based on the climate prediction model.

[0016] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:

[0017] By using the seasonal climate statistical prediction method provided in the embodiment of the present application, meteorological data within a specified time period before the determined year to be predicted (for example, 30 years before the year to be predicted) can be obtained, wherein the meteorological data includes monthly average precipitation data and monthly average temperature data for each month within the specified time period. The obtained meteorological data is preprocessed to convert the meteorological data into abnormal data, and the abnormal data is used as forecast object data, and the forecast object data is subjected to principal component decomposition to obtain forecast quantities of greater interest; next, variable data of each grid point in the global scope within the specified time period before the year to be predicted is obtained, and these variable data are converted into variable field data by processing these variable data; by respectively calculating the correlation coefficient between each forecast quantity and each global characteristic information field, a forecast factor is determined according to the correlation coefficient; finally, a distributed gradient boosting algorithm is used to statistically model the forecast factor to obtain a climate forecast model, and based on the climate forecast model, the climate of the year to be predicted is predicted. By adopting the seasonal climate statistical prediction method provided in the embodiment of the present application, meteorological data can be converted into abnormal meteorological data through preprocessing of meteorological data. Subsequent dynamic modeling based on the abnormal meteorological data can significantly improve the prediction skill of the prediction model in predicting the spatial distribution of climate anomalies; and by processing the variable data of grid points every month on a global scale, these variable data can be converted into feature information field data, thereby achieving feature enhancement of modeling features, thereby further improving the accuracy of meteorological prediction using the meteorological model. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 A schematic diagram of a specific process of a seasonal climate statistical prediction method provided in an embodiment of the present application;

[0020] Figure 2 A schematic diagram of the specific structure of a seasonal climate statistical prediction system provided in an embodiment of the present application;

[0021] Figure 3 A schematic diagram of the specific structure of a seasonal climate statistical prediction device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. 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.

[0023] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0024] The embodiments of the present application provide a seasonal climate statistical prediction method to address the problems of existing climate statistical prediction methods, such as the lack of targeted feature factor and model optimization, simple feature information and regression algorithms, and poor spatial distribution prediction skills for climate anomalies.

[0025] For ease of description, the following describes an implementation of the method using a seasonal climate statistical forecasting system as an example. It should be understood that using a seasonal climate statistical forecasting server as the execution subject of the method is merely an example and should not be construed as limiting the method.

[0026] The specific implementation flow chart of the seasonal climate statistical prediction method provided in this application is as follows Figure 1 As shown, it mainly includes the following steps:

[0027] Step 11, obtaining meteorological data within a specified time period according to the determined year to be predicted;

[0028] The meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period. It should be noted that since it is generally necessary to predict the domestic climate for a particular future season, the meteorological data obtained here is domestic meteorological data, which can be obtained from various domestic meteorological observation stations.

[0029] It should also be noted that the specified time period generally refers to several years before the year to be predicted, and the length is based on basically covering the adjacent interdecadal climate change cycle. In one embodiment, the specified time period can generally be 30 years before the year to be predicted.

[0030] Step 12: preprocessing the meteorological data obtained by executing step 11 to obtain forecast object data, and performing principal component decomposition on the forecast object data to obtain at least one forecast quantity;

[0031] Generally, when making climate forecasts for a future year, forecasts are often made for abnormal conditions in a certain season of that year, such as whether the precipitation in the summer of 2022 will be abnormal, or whether the precipitation in the autumn of 2022 will be abnormal. Therefore, in the embodiment of the present application, it is necessary to first convert the monthly average meteorological data within the specified time period obtained by executing step 11 into seasonal average data, and further determine the abnormal data in the historical data based on the seasonal average data.

[0032] In an embodiment of the present application, the method for preprocessing meteorological data may specifically include: calculating the seasonal meteorological data of each season within the specified time period based on the meteorological data for three consecutive months within the specified time period; calculating the seasonal average data of each season within the specified time period based on the seasonal meteorological data; determining abnormal data based on the seasonal meteorological data and the seasonal average data, and using the abnormal data as forecast object data.

[0033] Specifically, the monthly average data can be processed into seasonal data by adding the monthly average data for three consecutive months. Generally, March, April, and May are considered spring, June, July, and August are considered summer, September, October, and November are considered autumn, and December, January, and February of the following year are considered winter. The monthly average data for each historical year obtained by executing step 11 can be added up by month to obtain the corresponding seasonal data for each year. The average data for the four seasons over the past 30 years are then calculated to obtain seasonal average data. By comparing the seasonal data for each year over the past 30 years with the seasonal average data, abnormal data from the past 30 years can be determined and used as the forecast target data.

[0034] It should also be noted that temperature data is generally processed directly as temperature anomaly, which can be calculated by subtracting the seasonal temperature average from the seasonal temperature data. Precipitation data, on the other hand, needs to be processed as precipitation anomaly percentage, which can be calculated by dividing the difference between the seasonal precipitation data and the seasonal average precipitation data by the seasonal average precipitation data.

[0035] After the forecast object data for the past 30 years have been determined using the above method, it is necessary to perform principal component decomposition on the forecast object data, and then screen out at least one forecast quantity that users are more concerned about, so as to use it in subsequent statistical modeling.

[0036] In an embodiment of the present application, empirical orthogonal function analysis can be used to perform principal component decomposition on the forecast object data. In one embodiment, the method of performing principal component decomposition on the forecast object data to obtain the forecast quantity can specifically include: using the empirical orthogonal function analysis algorithm to perform principal component decomposition on the forecast object data to obtain the principal components and the explained variance corresponding to the principal components; selecting the principal component whose explained variance is greater than a preset threshold as the forecast quantity.

[0037] The empirical orthogonal function (EOF) method, also known as eigenvector analysis or principal component analysis (PCA), is a method for analyzing the structural features in matrix data and extracting the main data features.

[0038] Specifically, principal component decomposition can be performed according to the following sub-steps:

[0039] Sub-step 1201: Process the forecast object data into the form of deviations to obtain a data matrix Y m×n ;

[0040] Among them, m means that a single data has m dimensions, and n means that the time is upsampled n times.

[0041] Sub-step 1202: Calculate the cross product of Y and its transposed matrix to obtain a square matrix;

[0042] Specifically, the cross product of Y and its job transfer matrix can be calculated using the following formula [1]:

[0043]

[0044] Among them, if Y has been processed into deviation, then C is called the covariance matrix; if Y has been standardized (that is, the mean value of each row of data in C is 0 and the standard deviation is 1), then C is called the correlation coefficient matrix.

[0045] Sub-step 1203: Calculate the eigenvalues ​​(λ1, ..., m) and eigenvectors V of the square matrix C m×m , and the two satisfy the following formula [2]:

[0046] C m×m ×V m×m =V m×m ×∧ m×m [2]

[0047] Among them, ∧ is an m×m dimensional diagonal matrix, that is,

[0048]

[0049] Eigenvalues ​​are generally arranged from largest to smallest to reflect the importance of each principal component. Since the data Y is real observations, it should be greater than or equal to 0. The column of eigenvectors corresponding to each non-zero eigenvalue is the EOF. For example, the eigenvector corresponding to λ1 is called the first EOF mode, which is the first column of V, i.e., EOF1 = V(:,1). The eigenvector corresponding to the kth eigenvalue is the kth column of V, also called the kth mode.

[0050] Sub-step 1204: Calculate principal components.

[0051] By projecting EOF onto the original data matrix Y, we can obtain the time coefficients (i.e., principal components) corresponding to all spatial eigenvectors, as shown in the following formula [3]:

[0052]

[0053] Each row of data in PC is the time coefficient of each eigenvector. The first row is the time coefficient of the first EOF, and so on.

[0054] The above is the principal component (PC) of EOF obtained by calculating the data matrix Y. Therefore, the original data matrix Y can be completely restored using EOF and PC, as shown in the following formula [4]:

[0055] Y=EOF×PC [4]

[0056] In the embodiment of the present application, a principal component with an explained variance greater than 2% may be selected as a prediction variable.

[0057] Step 13, obtaining the designated variable data of the global grid points every month within the designated time period, performing time dimension averaging and difference preprocessing on the designated variable data, and obtaining at least one global range characteristic information field;

[0058] The specified variable data may include the following data:

[0059] Sea surface temperature data, sea ice data, sea level pressure data, 850hPa meridional wind data, 850hPa zonal wind data, 850hPa air temperature data, 500hPa geopotential height data, 200hPa zonal wind data, 200hPa meridional wind data, and 200hPa geopotential height data.

[0060] The above data can be obtained from the databases of the National Centers for Environmental Prediction (NCEP), the National Center for Atmospheric Research (NCAR), and the China Meteorological Administration (CRA), which store global gridded meteorological data from 1948 to the present.

[0061] The obtained variable data are preprocessed to convert them into feature information feature information fields to achieve feature enhancement. Specifically, in an embodiment of the present application, the specified variable data can be preprocessed according to the following method: for each type of specified variable data, the quarterly average value of each type of specified variable data is calculated based on the specified variable data of the adjacent three consecutive months within the specified time period; the quarterly turning point data signal of each type of specified variable data is calculated based on the data of the last month of the three consecutive months of specified variable data minus the average value of the data of the first two months; the quarterly average value and the quarterly turning point data signal are used as the global feature information field corresponding to the specified variable data.

[0062] Assuming that the sea surface temperature data for June, July and August 2021 need to be processed, the average value of the sea surface temperature data for these three months is first calculated as the continuity information data of the sea surface temperature data for that quarter; then the average value of the sea surface temperature in June and July is subtracted from the sea surface temperature data in August as the turning data signal for that quarter, and the continuity information data and turning information data are used as the characteristic information field of the sea surface temperature data in summer in 2021. After this processing, a single variable can be converted into a characteristic information field consisting of two variables: continuity information data and turning information data, so as to achieve feature enhancement of the variable data.

[0063] Step 14, using the leave-one-out method, respectively calculates the correlation coefficient between each forecast variable determined by executing step 12 and each global characteristic information field determined by executing step 13, and establishes a prediction factor based on the position of the maximum value of the calculated correlation coefficient;

[0064] In the embodiment of the present application, the correlation coefficient between each forecast quantity and each characteristic information field can be calculated according to the following formula [5]:

[0065]

[0066] Among them, pc represents the forecast amount, pc i represents the i-th forecast quantity, pc i is of length n year sequence, n year a quantity equal to the specified duration, Indicates pc i Remove the sequence of the jth value, j=(1,2,3……n year ), v represents the global feature information field, temp v It means removing the jth and the last value of v along the time dimension.

[0067] Based on the calculated correlation coefficients, the average correlation coefficient is calculated according to the following formula [6]:

[0068]

[0069] Among them, i var represents the i-th variable, i lat Indicates the i-th dimension in the latitude dimension lat Positions, i lon Indicates the i-th dimension in the longitude dimension lon locations.

[0070] Determine the absolute value of the average correlation coefficient corresponding to each global range characteristic information field, and determine the latitude and longitude positions where the absolute value of the average correlation coefficient corresponding to each global range characteristic information field is the largest according to the following formula [7]:

[0071] Location ACC =max(ACC) [7]

[0072] The latitude and longitude positions are determined as the predictor factors.

[0073] Step 15: Use a distributed gradient boosting algorithm to perform statistical modeling on the forecast factors and the forecast quantities determined by executing step 14 to obtain a climate forecast model, and predict the climate of the year to be predicted based on the climate forecast model.

[0074] In an embodiment of the present application, the XGboost algorithm may be used for statistical modeling.

[0075] In one embodiment, statistical modeling can be performed according to the following formula [8]:

[0076] PC i =XGboost(x) [8]

[0077] Here, x represents the prediction factor. It should be noted here that since the scheme of using the XGboost algorithm for statistical modeling is a common technical means in related fields, the detailed steps of using the XGboost algorithm for statistical modeling are not repeated here.

[0078] After completing the statistical modeling through the above steps, the climate of the predicted year can be predicted according to the following formula [9]:

[0079] Y i =XGboost(x(t=-1)) [9]

[0080] Among them, Y irepresents the forecast value of the i-th forecast quantity in the year to be forecasted. What is obtained at this time is one of the forecast principal components of the year to be forecasted. This principal component is obtained by the EOF method, and thus the principal component can also be restored to the original data matrix by the EOF method. Specifically, in the embodiment of the present application, the data can be restored according to the following formula

[10] to obtain the climate forecast result for the year to be forecasted:

[0081]

[0082] Among them, i pc represents the i-th principal component of the forecast quantity, Represents the spatial mode of the principal component of the i-th forecast quantity.

[0083] By using the climate prediction method provided in the embodiment of the present application, meteorological data within a specified time period before the determined year to be predicted (for example, 30 years before the year to be predicted) can be obtained, wherein the meteorological data includes the monthly average precipitation data and the monthly average temperature data for each month within the specified time period. The obtained meteorological data is preprocessed, the meteorological data is converted into abnormal data, and the abnormal data is used as the forecast object data, and the forecast object data is subjected to principal component decomposition to obtain a forecast quantity of greater interest; next, variable grid data of the three months before the reporting month on a global scale within the specified time period before the year to be predicted is obtained, and these variable data are converted into characteristic information field data by processing these variable data; by respectively calculating the correlation coefficient between each forecast quantity and each global characteristic information field, a forecast factor is determined according to the correlation coefficient; finally, a distributed gradient boosting algorithm is used to perform statistical modeling on the forecast factor and the forecast quantity to obtain a climate forecast model, and based on the climate forecast model, the climate of the year to be predicted is predicted. By using the climate prediction method provided in the embodiment of the present application, meteorological data can be converted into abnormal meteorological data through preprocessing of meteorological data. Subsequent dynamic modeling is performed based on the abnormal meteorological data, which can significantly improve the prediction skill of the prediction model in predicting the spatial distribution of climate anomalies; and by processing the variable data of grid points every month on a global scale, these variable data can be converted into feature information field data, thereby achieving feature enhancement of the modeling features, thereby further improving the accuracy of meteorological prediction using the meteorological model.

[0084] In one embodiment, the present application also provides a seasonal climate statistical prediction system to address the problems of existing climate statistical prediction methods, such as the lack of targeted feature factor and model optimization, simple feature information and regression algorithms, and poor spatial distribution prediction skills for climate anomalies. The specific structural diagram of the seasonal climate statistical prediction system is shown in the figure. Figure 2As shown, it includes: a meteorological data acquisition unit 21, a preprocessing unit 22, a characteristic information field determination unit 23, a forecast factor determination unit 24 and a prediction unit 25.

[0085] The meteorological data acquisition unit 21 is configured to acquire meteorological data within a specified time period according to the determined year to be predicted, wherein the meteorological data includes monthly average precipitation data and monthly average temperature data for each month within the specified time period;

[0086] a preprocessing unit 22 configured to preprocess the meteorological data to obtain forecast object data, and perform principal component decomposition on the forecast object data to obtain at least one forecast quantity;

[0087] The characteristic information field determining unit 23 is configured to obtain the designated variable data of the global grid points every month within the designated time period, and perform time-dimensional averaging and difference preprocessing on the designated variable data to obtain at least one global characteristic information field;

[0088] A prediction factor determination unit 24 is configured to calculate the correlation coefficients between the predicted quantities and the global characteristic information fields using a leave-one-out method, and to establish a prediction factor based on the maximum value position of the correlation coefficient;

[0089] The prediction unit 25 is used to use a distributed gradient boosting algorithm to perform statistical modeling on the prediction factors to obtain a climate prediction model, and predict the climate of the year to be predicted based on the climate prediction model.

[0090] In one embodiment, the preprocessing unit 22 is specifically used to calculate the seasonal meteorological data of each season within the specified time period based on the meteorological data of three consecutive months within the specified time period; calculate the seasonal average data of each season within the specified time period based on the seasonal meteorological data; determine abnormal data based on the seasonal meteorological data and the seasonal average data, and use the abnormal data as forecast object data.

[0091] In one embodiment, the preprocessing unit 22 is specifically used to: use the empirical orthogonal function analysis algorithm to perform principal component decomposition on the forecast object data to obtain principal components and the explained variances corresponding to the principal components; select the principal components whose explained variances are greater than a preset threshold as the forecast quantity.

[0092] In one embodiment, the specified variable data specifically includes: sea surface temperature data, sea ice data, sea level pressure data, 850hPa meridional wind data, 850hPa zonal wind data, 850hPa air temperature data, 500hPa geopotential height data, 200hPa zonal wind data, 200hPa meridional wind data and 200hPa geopotential height data.

[0093] In one embodiment, the characteristic information field determination unit 23 is specifically used to: for each type of specified variable data, calculate the quarterly average value of each type of specified variable data based on the specified variable data for three consecutive months within the specified time period; calculate the quarterly turning data signal of each type of specified variable data based on the average value of the data of the first two months minus the data of the last month of the three consecutive months of specified variable data; and use the quarterly average value and the quarterly turning data signal as the global characteristic information field corresponding to the specified variable data.

[0094] In one embodiment, the prediction factor determination unit 24 is specifically configured to calculate the correlation coefficient between each of the prediction quantities and each of the global characteristic information fields according to the following formula:

[0095]

[0096] Among them, pc represents the forecast amount, pc i represents the i-th forecast quantity, pc i is of length n year sequence, n year a quantity equal to the specified duration, Indicates pc i Remove the sequence of the jth value, j=(1,2,3……n year ), v represents the global feature information field, temp v It means that v removes the jth and the last value along the time dimension;

[0097] Based on the correlation coefficients, the average correlation coefficient is calculated according to the following formula:

[0098]

[0099] Among them, i var represents the i-th variable, i lat Indicates the i-th dimension in the latitude dimension lat Positions, i lon Indicates the i-th dimension in the longitude dimension lon positions;

[0100] Determining the absolute value of the average correlation coefficient corresponding to each of the global range characteristic information fields;

[0101] The latitude and longitude positions at which the absolute value of the average correlation coefficient corresponding to each of the global range characteristic information fields is the largest are determined according to the following formula:

[0102] Location ACC =max(|ACC|)

[0103] The latitude and longitude positions are determined as the predictor factors.

[0104] In one embodiment, the prediction unit 25 is specifically configured to perform statistical modeling according to the following formula:

[0105] PC i =XGboost(x)

[0106] Where x represents the predictor;

[0107] The climate of the year to be predicted is predicted according to the following formula:

[0108] Y i =XGboost(x(t=-1))

[0109] Among them, Y i It represents the forecast value of the i-th forecast quantity in the forecast year;

[0110] The forecast quantity of the forecast year is calculated according to the following formula to obtain the climate forecast result of the forecast year:

[0111]

[0112] Among them, i pc represents the i-th principal component of the forecast quantity, Represents the spatial mode of the principal component of the i-th forecast quantity.

[0113] By using the seasonal climate statistical prediction system provided in the embodiment of the present application, meteorological data within a specified time period before the determined year to be predicted (for example, 30 years before the year to be predicted) can be obtained, wherein the meteorological data includes monthly average precipitation data and monthly average temperature data for each month within the specified time period. The obtained meteorological data is preprocessed to convert the meteorological data into abnormal data, and the abnormal data is used as forecast object data, and the forecast object data is subjected to principal component decomposition to obtain forecast quantities of greater interest; next, variable data of each grid point on a global scale within a specified time period before the year to be predicted is obtained, and these variable data are converted into characteristic information field data by processing these variable data; by respectively calculating the correlation coefficient between each forecast quantity and each global characteristic information field, a forecast factor is determined based on the correlation coefficient; finally, a distributed gradient boosting algorithm is used to statistically model the forecast factor and the forecast quantity to obtain a climate forecast model, and based on the climate forecast model, the climate of the year to be predicted is predicted. By using the climate prediction method provided in the embodiment of the present application, meteorological data can be converted into abnormal meteorological data through preprocessing of meteorological data. Subsequent dynamic modeling is performed based on the abnormal meteorological data, which can significantly improve the prediction skill of the prediction model in predicting the spatial distribution of climate anomalies; and by processing the variable data of grid points every month on a global scale, these variable data can be converted into feature information field data, thereby achieving feature enhancement of the modeling features, thereby further improving the accuracy of meteorological prediction using the meteorological model.

[0114] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 3 At the hardware level, the electronic device includes a processor and, optionally, an internal bus, a network interface, and memory. The memory may include internal memory, such as high-speed random-access memory (RAM), or non-volatile memory, such as at least one disk drive. Of course, the electronic device may also include other hardware required for its services.

[0115] The processor, network interface, and memory can be interconnected via an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0116] The memory is used to store programs. Specifically, the program may include program code, which includes computer operating instructions. The memory may include internal memory and non-volatile memory, and provides instructions and data to the processor.

[0117] The processor reads the corresponding computer program from the non-volatile memory into the internal memory and then runs it, forming a data synchronization device at the logical level. The processor executes the program stored in the memory and is specifically configured to perform the following operations: based on the determined year to be predicted, obtain meteorological data within a specified time period, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; preprocess the meteorological data to obtain forecast target data, and perform principal component decomposition on the forecast target data to obtain at least one forecast quantity; obtain designated variable data for each month at a global grid point within the specified time period, perform time-dimensional averaging and difference preprocessing on the designated variable data to obtain at least one global-scale characteristic information field; calculate the correlation coefficient between each forecast quantity and each global-scale characteristic information field using the leave-one-out method, and establish a forecast factor based on the location of the maximum value of the correlation coefficient; perform statistical modeling on the forecast factor using a distributed gradient boosting algorithm to obtain a climate prediction model; and predict the climate for the year to be predicted based on the climate prediction model.

[0118] The above application Figure 3The methods performed by the electronic device for seasonal climate statistical forecasting disclosed in the illustrated embodiments can be applied to or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be performed by hardware integrated logic circuits in the processor or by software instructions. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The methods, steps, and logic block diagrams disclosed in the embodiments of this application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method.

[0119] Of course, in addition to software implementation, the electronic device of this application does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0120] The embodiment of the present application also provides a computer-readable storage medium, which stores one or more programs, wherein the one or more programs include instructions, which, when executed by a portable electronic device including multiple application programs, can enable the portable electronic device to execute Figure 1 The method of the embodiment shown is specifically used to perform the following operations:

[0121] According to the determined year to be predicted, meteorological data within a specified time period is obtained, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; the meteorological data is preprocessed to obtain forecast object data, and the forecast object data is decomposed into principal components to obtain at least one forecast quantity; designated variable data of each month at a global grid point within the specified time period is obtained, and the designated variable data is preprocessed in terms of time dimension average and difference to obtain at least one global-scale characteristic information field; the correlation coefficient between each forecast quantity and each global-scale characteristic information field is calculated using the leave-one-out method, and a forecast factor is established according to the position of the maximum value of the correlation coefficient; the forecast factor is statistically modeled using a distributed gradient boosting algorithm to obtain a climate prediction model, and the climate of the year to be predicted is predicted based on the climate prediction model.

[0122] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0123] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, 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 flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0124] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0125] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0126] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0127] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0128] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0129] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a..." does not preclude the presence of additional identical elements in the process, method, commodity, or apparatus that includes the element.

[0130] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0131] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A seasonal climate statistical prediction method, characterized in that: include: Acquiring meteorological data within a specified time period according to the determined year to be predicted, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; Preprocessing the meteorological data to obtain forecast object data, and performing principal component decomposition on the forecast object data to obtain at least one forecast quantity; Obtaining designated variable data for each global grid point per month within the designated time period, performing time-dimensional averaging and differential preprocessing on the designated variable data to obtain at least one global characteristic information field, wherein the designated variable data specifically include: sea surface temperature data, sea ice data, sea level pressure data, 850hPa meridional wind data, 850hPa zonal wind data, 850hPa air temperature data, 500hPa geopotential height data, 200hPa zonal wind data, 200hPa meridional wind data, and 200hPa geopotential height data; Calculating the correlation coefficients between the forecast quantities and the global characteristic information fields respectively using a leave-one-out method, and establishing a forecast factor based on the maximum value position of the correlation coefficient; Using a distributed gradient boosting algorithm, statistically modeling the forecast factors to obtain a climate prediction model, and predicting the climate of the year to be predicted based on the climate prediction model; Obtaining the designated variable data for each month at the global grid point within the designated time period, performing time-dimensional averaging and difference preprocessing on the designated variable data to obtain at least one global characteristic information field, specifically including: For each type of designated variable data, calculate the quarterly average value of each type of designated variable data based on the designated variable data for three consecutive months within the designated period; Calculate the quarterly turning point information data of each type of specified variable data by subtracting the average of the data of the first two months from the data of the last month of the specified variable data for three consecutive months; Using the quarterly average value and the quarterly turning point information data as the global feature information field corresponding to the designated variable data; The preprocessing of the meteorological data to obtain forecast object data specifically includes: Calculate seasonal meteorological data for each season within the specified time period based on meteorological data for three consecutive months within the specified time period; Calculating seasonal average data for each season within the specified time period based on the seasonal meteorological data; Abnormal data is determined based on the seasonal meteorological data and the seasonal average data, and the abnormal data is used as forecast target data.

2. The method according to claim 1, characterized in that The performing principal component decomposition on the forecast object data to obtain at least one forecast quantity specifically includes: Using an empirical orthogonal function analysis algorithm, the forecast object data is subjected to principal component decomposition to obtain principal components and explained variances corresponding to the principal components; The principal component with explained variance greater than the preset threshold is selected as the predicted variable.

3. The method according to claim 1, characterized in that The method of using the leave-one-out method to calculate the correlation coefficients between the respective forecast quantities and the respective global characteristic information fields, and establishing the forecast factor according to the maximum position of the correlation coefficient, specifically includes: The correlation coefficients between the forecast quantities and the global characteristic information fields are calculated according to the following formula: Among them, PC represents the forecast quantity, PC i represents the i-th forecast quantity, PC i is of length n year sequence, n year a quantity equal to the specified duration, Indicates PC i Remove the sequence of the jth value, j=(1,2,3……n year ), v represents the global feature information field, It means that v removes the jth and the last value along the time dimension; Based on the correlation coefficients, the average correlation coefficient is calculated according to the following formula: in, represents the i-th variable, Indicates the latitude dimension positions, Indicates the first degree in longitude positions; Determining the absolute value of the average correlation coefficient corresponding to each of the global range characteristic information fields; The latitude and longitude positions at which the absolute value of the average correlation coefficient corresponding to each of the global range characteristic information fields is the largest are determined according to the following formula: The characteristic information field of the latitude and longitude position is determined as the prediction factor.

4. The method according to claim 1, wherein Using a distributed gradient boosting algorithm, statistical modeling is performed on the forecast factors to obtain a climate prediction model. Based on the climate prediction model, the climate of the year to be predicted is predicted, specifically including: Statistical modeling was performed according to the following formula: Where x represents the predictor; The climate of the year to be predicted is predicted according to the following formula: Among them, Y i It represents the forecast value of the i-th forecast quantity in the forecast year; The forecast quantity of the forecast year is calculated according to the following formula to obtain the climate forecast result of the forecast year: Among them, i pc represents the i-th principal component of the forecast quantity, Represents the spatial mode of the principal component of the i-th forecast quantity.

5. A seasonal climate statistical prediction system, characterized in that: include: A meteorological data acquisition unit, configured to acquire meteorological data within a specified time period according to the determined year to be predicted, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; a preprocessing unit, configured to preprocess the meteorological data to obtain forecast object data, and perform principal component decomposition on the forecast object data to obtain at least one forecast quantity; a characteristic information field determining unit, configured to obtain designated variable data for each month at a global grid point within the designated time period, perform time-dimensional averaging and differential preprocessing on the designated variable data, and obtain at least one global characteristic information field, wherein the designated variable data specifically include: sea surface temperature data, sea ice data, sea level pressure data, 850hPa meridional wind data, 850hPa zonal wind data, 850hPa air temperature data, 500hPa geopotential height data, 200hPa zonal wind data, 200hPa meridional wind data, and 200hPa geopotential height data; A prediction factor determination unit, configured to calculate the correlation coefficients between the respective predicted quantities and the respective global characteristic information fields using a leave-one-out method, and establish a prediction factor based on the position of the maximum value of the correlation coefficient; A prediction unit is configured to perform statistical modeling on the prediction factors using a distributed gradient boosting algorithm to obtain a climate prediction model, and predict the climate of the year to be predicted based on the climate prediction model; The characteristic information field determining unit is configured to: For each type of designated variable data, calculate the quarterly average value of each type of designated variable data based on the designated variable data for three consecutive months within the designated period; Calculate the quarterly turning point information data of each type of specified variable data by subtracting the average of the data of the first two months from the data of the last month of the specified variable data for three consecutive months; Using the quarterly average value and the quarterly turning point information data as the global feature information field corresponding to the designated variable data; The pre-processing unit is used to: The preprocessing of the meteorological data to obtain forecast object data specifically includes: Calculate seasonal meteorological data for each season within the specified time period based on meteorological data for three consecutive months within the specified time period; Calculating seasonal average data for each season within the specified time period based on the seasonal meteorological data; Abnormal data is determined based on the seasonal meteorological data and the seasonal average data, and the abnormal data is used as forecast target data.

6. A seasonal climate statistical prediction device comprising: processor; and a memory arranged to store computer-executable instructions that, when executed, cause the processor to: Acquiring meteorological data within a specified time period according to the determined year to be predicted, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; Preprocessing the meteorological data to obtain forecast object data, and performing principal component decomposition on the forecast object data to obtain at least one forecast quantity; Obtaining designated variable data for each global grid point per month within the designated time period, performing time-dimensional averaging and differential preprocessing on the designated variable data to obtain at least one global characteristic information field, wherein the designated variable data specifically include: sea surface temperature data, sea ice data, sea level pressure data, 850hPa meridional wind data, 850hPa zonal wind data, 850hPa air temperature data, 500hPa geopotential height data, 200hPa zonal wind data, 200hPa meridional wind data, and 200hPa geopotential height data; Calculating the correlation coefficients between the forecast quantities and the global characteristic information fields respectively using a leave-one-out method, and establishing a forecast factor based on the maximum value position of the correlation coefficient; Using a distributed gradient boosting algorithm, statistically modeling the forecast factors to obtain a climate prediction model, and predicting the climate of the year to be predicted based on the climate prediction model; Obtaining the designated variable data for each month at the global grid point within the designated time period, performing time-dimensional averaging and difference preprocessing on the designated variable data to obtain at least one global characteristic information field, specifically including: For each type of designated variable data, calculate the quarterly average value of each type of designated variable data based on the designated variable data for three consecutive months within the designated period; Calculate the quarterly turning point information data of each type of specified variable data by subtracting the average of the data of the first two months from the data of the last month of the specified variable data for three consecutive months; Using the quarterly average value and the quarterly turning point information data as the global feature information field corresponding to the designated variable data; The preprocessing of the meteorological data to obtain forecast object data specifically includes: Calculate seasonal meteorological data for each season within the specified time period based on meteorological data for three consecutive months within the specified time period; Calculating seasonal average data for each season within the specified time period based on the seasonal meteorological data; Abnormal data is determined based on the seasonal meteorological data and the seasonal average data, and the abnormal data is used as forecast target data.

7. A computer-readable storage medium storing one or more programs that, when executed by an electronic device including a plurality of application programs, causes the electronic device to perform the following operations: Acquiring meteorological data within a specified time period according to the determined year to be predicted, wherein the meteorological data includes monthly average precipitation data or monthly average temperature data for each month within the specified time period; Preprocessing the meteorological data to obtain forecast object data, and performing principal component decomposition on the forecast object data to obtain at least one forecast quantity; Obtaining designated variable data for each global grid point per month within the designated time period, performing time-dimensional averaging and differential preprocessing on the designated variable data to obtain at least one global characteristic information field, wherein the designated variable data specifically include: sea surface temperature data, sea ice data, sea level pressure data, 850hPa meridional wind data, 850hPa zonal wind data, 850hPa air temperature data, 500hPa geopotential height data, 200hPa zonal wind data, 200hPa meridional wind data, and 200hPa geopotential height data; respectively calculating correlation coefficients between the forecast quantities and the global characteristic information fields, and establishing forecast factors according to the positions of the maximum values ​​of the correlation coefficients; Using a distributed gradient boosting algorithm, statistically modeling the forecast factors to obtain a climate prediction model, and predicting the climate of the year to be predicted based on the climate prediction model; Obtaining the designated variable data for each month at the global grid point within the designated time period, performing time-dimensional averaging and difference preprocessing on the designated variable data to obtain at least one global characteristic information field, specifically including: For each type of designated variable data, calculate the quarterly average value of each type of designated variable data based on the designated variable data for three consecutive months within the designated period; Calculate the quarterly turning point information data of each type of specified variable data by subtracting the average of the data of the first two months from the data of the last month of the specified variable data for three consecutive months; Using the quarterly average value and the quarterly turning point information data as the global feature information field corresponding to the designated variable data; The preprocessing of the meteorological data to obtain forecast object data specifically includes: Calculate seasonal meteorological data for each season within the specified time period based on meteorological data for three consecutive months within the specified time period; Calculating seasonal average data for each season within the specified time period based on the seasonal meteorological data; Abnormal data is determined based on the seasonal meteorological data and the seasonal average data, and the abnormal data is used as forecast target data.

Images