A spatial probability analysis method and system based on the corresponding relationship between precipitation forecast and teleconnection

Abstract

The present invention relates to the technical field of precipitation forecast analysis, and specifically to a spatial probability analysis method and system based on the correspondence between precipitation forecasts and teleconnections. The method comprises the following steps: obtaining a sample sequence of precipitation forecasts to be analyzed, as well as corresponding sample sequences of observed precipitation and meteorological factors; calculating the forecast-observation correlation coefficient and meteorological factor-observed precipitation teleconnection coefficient for each grid based on the obtained sample sequence, and classifying each grid based on the significance of the forecast-observation correlation coefficient and meteorological factor-observed precipitation teleconnection coefficient; determining the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results; calculating spatial weight coefficients based on the spatial coordinates of the grids to obtain a spatial weight coefficient matrix; and calculating the spatial consistency probability that the forecast-observation correlation coefficient is significantly positively correlated based on the spatial weight coefficient matrix and the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient.

Description

Technical Field

[0001] The present invention relates to the technical field of precipitation forecast analysis, and more specifically, to a spatial probability analysis method and system based on the corresponding relationship between precipitation forecast and teleconnection. Background Art

[0002] Scientific and accurate seasonal precipitation forecasts have important application value in flood prevention and disaster reduction, flood resource utilization and reservoir operation. The Earth's Eleventh Annual Oscillation (ENSO) phenomenon plays an important role in indicative of global seasonal precipitation. Extensive observational data and analytical studies have demonstrated that ENSO, through atmospheric teleconnections, can significantly influence regional and even global precipitation. Consequently, operational forecast centers in some countries and regions specialize in observing and forecasting ENSO events, using this information to provide statistical forecasts of future precipitation anomalies. Furthermore, global climate models (GCMs) have steadily developed in recent years, providing a wealth of meteorological driving data on precipitation and temperature. These forecast data, with considerable accuracy and a long forecast horizon, are increasingly being incorporated into operational precipitation forecasts.

[0003] The understanding and research of ENSO and precipitation teleconnection coefficients provide important support for global seasonal precipitation forecasts. Many analytical studies have shown that the main source of predictability information for seasonal precipitation forecasts is the ENSO signal, and pointed out the correspondence between the strength of the ENSO and regional precipitation teleconnection coefficients and the accuracy of precipitation forecasts. Therefore, the ability of global meteorological models to capture and characterize the ENSO-precipitation teleconnection coefficient provides a key entry point for assessing the applicability of global seasonal precipitation forecasts. In actual precipitation forecast assessment and analysis, direct comparisons are often made between forecast accuracy and the similarity of the spatial distribution of teleconnection intensity, making it difficult to provide a quantitative description of the corresponding relationship between the two. In addition, the correlation coefficients of adjacent regions are usually not independent, but rather have a strong correlation. Traditional precipitation forecast assessment and analysis methods generally target a single grid, ignoring the spatial properties of the variables, resulting in an inability to accurately assess the accuracy of precipitation forecasts. Summary of the Invention

[0004] In order to overcome the defects of the above-mentioned existing precipitation forecast evaluation and analysis, such as the difficulty in giving a quantitative description of the forecast accuracy and the spatial distribution of teleconnection intensity, and the neglect of the spatial attributes of variables, which leads to the inability to accurately evaluate the precipitation forecast accuracy, the present invention provides a spatial probability analysis method and system based on the corresponding relationship between precipitation forecast and teleconnection.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0006] A spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection includes the following steps:

[0007] Obtain the sample sequence of precipitation forecasts to be analyzed, as well as the corresponding sample sequences of observed precipitation and meteorological factors;

[0008] Based on the obtained sample sequence, the forecast-observation correlation coefficient and meteorological factor-observed precipitation teleconnection coefficient of each grid are calculated respectively, and each grid is classified according to the significance of the forecast-observation correlation coefficient and meteorological factor-observed precipitation teleconnection coefficient;

[0009] Determine the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results;

[0010] Calculate the spatial weight coefficient according to the spatial coordinates of the grid to obtain the spatial weight coefficient matrix;

[0011] Based on the spatial weight coefficient matrix and the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient, the spatial consistency probability of the forecast-observation correlation coefficient being a significant positive correlation is calculated.

[0012] Furthermore, the present invention proposes a spatial probability analysis system based on the correspondence between precipitation forecasts and teleconnections, which applies the above-mentioned spatial probability analysis method based on the correspondence between precipitation forecasts and teleconnections. The system includes a data acquisition module, a correlation coefficient calculation module, a classification module, a significance determination module, a spatial weight coefficient calculation module, and a spatial consistency probability analysis module.

[0013] In this technical solution, the data acquisition module is used to obtain the sample sequence of precipitation forecast to be analyzed, as well as the corresponding sample sequence of observed precipitation and meteorological factors; the correlation coefficient calculation module is used to calculate the forecast-observation correlation coefficient and the meteorological factor-observed precipitation telecorrelation coefficient of each grid in the target area according to the acquired sample sequence; the classification module is used to analyze the significance of the forecast-observation correlation coefficient and the meteorological factor-observed precipitation telecorrelation coefficient, and classify each grid according to the analysis results; the significance judgment module is used to judge the correspondence between the forecast-observation correlation coefficient and the telecorrelation coefficient based on the grid classification results; the spatial weight coefficient calculation module is used to calculate the spatial weight coefficient according to the spatial coordinates of the grid to obtain the spatial weight coefficient matrix; the spatial consistency probability analysis module is used to calculate the spatial consistency probability that the forecast-observation correlation coefficient is significantly positively correlated, and the spatial consistency probability of the corresponding relationship between the forecast-observation correlation coefficient and different telecorrelation coefficients according to the spatial weight coefficient matrix and the correspondence between the forecast-observation correlation coefficient and the telecorrelation coefficient.

[0014] Compared with the existing technology, the beneficial effect of the technical solution of the present invention is: by combining the spatial relationship and probability of predicted precipitation, the present invention quantifies the spatial consistency probability of the forecast and observation correlation coefficient being significantly positive, and can decompose it into spatial consistency probabilities with different corresponding relationships with teleconnection effects, thereby providing a reference for the evaluation and selection of precipitation forecast products. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of a spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to an embodiment of the present invention.

[0016] Figure 2 Schematic diagram of the DJF forecast-observation correlation coefficient significance classification results.

[0017] Figure 3 Schematic diagram of the significance classification results of DJF's teleconnection coefficient.

[0018] Figure 4 This is the spatial consistency probability distribution map of the forecast-observation correlation coefficient being significantly positive.

[0019] Figure 5 This is the spatial consistency probability distribution map where both the forecast-observation correlation coefficient and the teleconnection coefficient are significantly positively correlated.

[0020] Figure 6 This is the spatial consistency probability distribution map with a significant positive correlation between the forecast and observation correlation coefficient and an insignificant teleconnection coefficient.

[0021] Figure 7 This is the spatial consistency probability distribution map where the forecast-observation relationship is significantly positive and the teleconnection coefficient is significantly negative.

[0022] Figure 8 4 is an architecture diagram of a spatial probability analysis system based on the correspondence between precipitation forecast and teleconnection according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The accompanying drawings are for illustrative purposes only and are not to be construed as limiting this patent;

[0024] It is understandable to those skilled in the art that some well-known structures and descriptions thereof may be omitted in the drawings.

[0025] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0026] Example 1

[0027] This embodiment proposes a spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection, such as Figure 1, which is a flow chart of the spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection in this embodiment.

[0028] The spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection proposed in this embodiment includes the following steps:

[0029] S1. Obtain the sample sequence of precipitation forecasts to be analyzed, as well as the corresponding sample sequences of observed precipitation and meteorological factors.

[0030] S2. Based on the obtained sample sequence, calculate the forecast-observation correlation coefficient and meteorological factor-observation precipitation teleconnection coefficient of each grid respectively, and classify each grid according to the significance of the forecast-observation correlation coefficient and meteorological factor-observation precipitation teleconnection coefficient.

[0031] S3. Determine the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results.

[0032] S4. Calculate the spatial weight coefficient according to the spatial coordinates of the grid to obtain a spatial weight coefficient matrix.

[0033] S5. Calculate the spatial consistency probability that the forecast-observation correlation coefficient is significantly positively correlated based on the spatial weight coefficient matrix and the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient.

[0034] In this embodiment, by combining the spatial relationship and probability of predicted precipitation, the spatial consistency probability of a significantly positive correlation coefficient between the forecast and observation is quantified, and it can be decomposed into spatial consistency probabilities corresponding to different teleconnection effects, thereby providing a reference for the evaluation and selection of precipitation forecast products.

[0035] In an optional embodiment, the step of calculating the forecast-observation correlation coefficient and the meteorological factor-observation precipitation teleconnection coefficient of each grid respectively based on the acquired sample sequence includes:

[0036] S2.1. Extract the forecast precipitation data and observed precipitation data of the target area grid based on the obtained sample sequence.

[0037] S2.2. Calculate the forecast-observation correlation coefficient r(o,f) on a grid-by-grid basis; its expression is as follows:

[0038]

[0039] In the formula, o k represents the observed precipitation data for the kth year, f k represents the forecast precipitation data for the kth year; They represent the mean of historical observed precipitation data and the mean of historical forecast precipitation data respectively.

[0040] S2.3. Calculate the meteorological factor-observed precipitation teleconnection coefficient r(o,η) for each grid. Its expression is as follows:

[0041]

[0042] Where η k represents the meteorological factor index of the kth year, Represents the mean of historical meteorological factor index.

[0043] Among them, the average historical precipitation data It is calculated based on the sample sequence of historical precipitation observations, and its expression is as follows:

[0044]

[0045] Where K is the total number of years in the historical sample sequence.

[0046] Similarly, the mean of historical precipitation forecast data It is calculated based on the sample sequence of historical precipitation forecasts, and its expression is as follows:

[0047]

[0048] In this embodiment, for the El Nino-Southern Oscillation (El For teleconnection, the commonly used meteorological factors are index.

[0049] Furthermore, the steps of classifying each grid according to the significance of the forecast-observation correlation coefficient and the meteorological factor-observation precipitation telecorrelation coefficient include:

[0050] S2.4. Based on the preset significance level α, as well as the forecast-observation correlation coefficient and meteorological factor-observation precipitation telecorrelation coefficient of each grid, perform significance judgment and classification for each grid in the target area.

[0051] The probability density function of the correlation coefficient r is estimated using the Beta function:

[0052]

[0053] Where r is the correlation coefficient, B is the Beta function, and n is the number of forecast and observation samples used to calculate the correlation coefficient. The cumulative probability density function corresponding to this probability density function is denoted as F. Then, at the significance level α, the 100×(1-α / 2) quantile of the correlation coefficient r is expressed as r 1-α / 2 :

[0054] r 1-α / 2 =F -1(1-α / 2)

[0055] The 100×(α / 2) quantile of the correlation coefficient r is expressed as r α / 2 :

[0056] r α / 2 =F -1 (α / 2).

[0057] Then for the forecast-observation correlation coefficient r(o,f):

[0058] If the forecast-observation correlation coefficient r(o,f) is greater than r 1-α / 2 , it is judged to be significantly positively correlated;

[0059] If the forecast-observation correlation coefficient r(o,f) is less than or equal to r 1-α / 2 , and greater than r α / 2 , it is judged as not significant;

[0060] If the forecast-observation correlation coefficient r(o,f) is less than r α / 2 , it is judged to be a significant negative correlation.

[0061] For meteorological factor-observation precipitation teleconnection coefficient r(o,η):

[0062] If the teleconnection coefficient r(o,η) is greater than r 1-α / 2 , it is judged to be significantly positively correlated;

[0063] If the teleconnection coefficient r(o,η) is less than or equal to r 1-α / 2 , and greater than r α / 2 , it is judged as not significant;

[0064] If the teleconnection coefficient r(o,η) is less than r α / 2 , it is judged to be a significant negative correlation.

[0065] Among them, under a given significance level α, each grid is divided into three categories according to the significance of its correlation coefficient: significantly positive correlation (significantly positive, P), non-significant correlation (non-significant, ns) and significantly negative correlation (significantly negative, N).

[0066] That is, the forecast-observation correlation coefficient r(o,f) can be divided into three categories:

[0067]

[0068] The meteorological factor-observation precipitation teleconnection coefficient r(o,η) can be divided into three categories:

[0069]

[0070] Furthermore, the step of judging the correspondence between the forecast-observation correlation coefficient and the telecorrelation coefficient according to the grid classification result includes: judging grid by grid whether the forecast-observation correlation is significantly positively correlated and the meteorological factor-observed precipitation telecorrelation is significantly positively correlated, the forecast-observation correlation is significantly positively correlated and the meteorological factor-observed precipitation telecorrelation is not significant, or the forecast-observation correlation is significantly positively correlated and the meteorological factor-observed precipitation telecorrelation is significantly negatively correlated, and constructing a corresponding relationship vector through Boolean numbers.

[0071] This embodiment first determines the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient when the forecast-observation correlation coefficient r(o,f) is significantly positively correlated, and constructs a correspondence vector using Boolean numbers. The expression is as follows:

[0072] b(P AC &P ENSO )=[x i ] N×1

[0073] b(P AC &ns ENSO )=[x i ] N×1

[0074] b(P AC &N ENSO )=[x i ] N×1

[0075] Where N is the total number of grids in the target area.

[0076] b(P AC &P ENSO ) is a significant positive correlation between forecast and observation. AC The meteorological factor-observed precipitation telecorrelation is significantly positively correlated P ENSO Boolean vector, when the grid i satisfies P AC &P ENSO , then x i The value is 1, otherwise x i The value is 0.

[0077] b(P AC &ns ENSO ) indicates that the forecast-observation correlation is a significant positive correlation P AC The meteorological factor-observed precipitation telecorrelation is not significant. ENSO Boolean vector, when the grid i satisfies P AC &ns ENSO , then x i The value is 1, otherwise xi The value is 0.

[0078] b(P AC &N ENSO ) indicates that the forecast-observation correlation is a significant positive correlation P AC The meteorological factor-observed precipitation telecorrelation is significantly negatively correlated. ENSO Boolean vector, when the grid i satisfies P AC &N ENSO , then x i The value is 1, otherwise x i The value is 0.

[0079] In an optional embodiment, the step of calculating the spatial weight coefficient according to the spatial coordinates of the grid and obtaining the spatial weight coefficient matrix includes:

[0080] S4.1. Mark the coordinates of the target area grid with the upper left corner of the target area as the origin.

[0081] S4.2. Calculate the spatial weight coefficient for any two grid coordinates using the quadratic attenuation function of distance. The expression is as follows:

[0082]

[0083]

[0084]

[0085] Where, d ij is the grid point (u i ,v i ) and grid points (u j ,v j d is the weight coefficient bandwidth value, and optionally, the weight coefficient bandwidth value d is 5.

[0086] S4.3. Construct a spatial weight coefficient matrix based on the spatial weight coefficients of any two grid coordinates. Its expression is as follows:

[0087] W=[w ij ] N×N

[0088] Where N is the total number of grids in the target area.

[0089] Furthermore, the spatial weight coefficient matrix A is subjected to row normalization, and each spatial weight coefficient is subjected to row normalization; its expression is as follows:

[0090]

[0091] This embodiment performs row normalization on each weight coefficient to ensure that the sum of the weight coefficients of each row is equal to 1.

[0092] In an optional embodiment, the step of calculating the spatial consistency probability that the forecast-observation correlation coefficient is significantly positively correlated according to the spatial weight coefficient matrix A and the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient includes:

[0093] Based on the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient of each grid in the target area, the Boolean vector of the forecast-observation correlation coefficient of each grid is calculated to be significantly positively correlated. Then, the Boolean vector is multiplied by the spatial weight coefficient corresponding to the grid to calculate the spatial consistency probability that the forecast-observation correlation coefficient of the corresponding grid is significantly positively correlated. The expression is as follows:

[0094] P(P AC )=A·b(P AC )=[p i ] N×1

[0095] Where, b(P AC ) is a Boolean vector of grid forecast-observation correlation coefficients that are significantly positive, p i Indicates the spatial consistency probability that the forecast-observation correlation coefficient for grid i is significantly positive.

[0096] The forecast-observation correlation coefficient is a Boolean vector b(P AC ), including the Boolean vector b(P AC &P ENSO ), the Boolean vector b(P AC &ns ENSO ), a Boolean vector b(P AC &N ENSO ).

[0097] The forecast-observation correlation coefficient calculated from this is the spatial consistency probability P(P AC ), including the spatial consistency probability P(P) of the corresponding grid belonging to both the forecast-observation correlation and the meteorological factor-observation precipitation teleconnection. AC &P ENSO ), the probability of spatial consistency P (P) that the forecast-observation correlation is significantly positive and the meteorological factor-observation precipitation telecorrelation is insignificant AC&ns ENSO ), and the spatial consistency probability P(P AC &N ENSO ). Its expression is as follows:

[0098] P(P AC )=P(P AC &P ENSO )+P(P AC &ns ENSO )+P(P AC &N ENSO )

[0099] =A·b(P AC &P ENSO )+A·b(P AC &ns ENSO )+A·b(P AC &N ENSO )

[0100] =[p i ] N×1

[0101] Optionally, similarly, this embodiment can calculate the spatial consistency probability P (ns) that the forecast-observation correlation coefficient is insignificant based on the spatial weight coefficient matrix A and the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient. AC ) and the spatial consistency probability P(N AC ), further providing a reference for the evaluation and selection of precipitation forecast products.

[0102] Example 2

[0103] This embodiment tests the spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection proposed in Example 1.

[0104] In this embodiment, the monthly precipitation data of the Climate Prediction Center global daily Unified Raingauge Database (CPC-URD) of the National Oceanic and Atmosphere Administration (NOAA) from 1982 to 2010 are used as observation data, and the observed precipitation for three consecutive months is accumulated to obtain seasonal precipitation; the second-generation climate forecast system CFSv2 of the Centers for Environmental Prediction (NCEP) is used as forecast precipitation data; The index represents the ENSO phenomenon.

[0105] CFSv2 uses seasonal precipitation forecasts for a zero-month forecast period. For example, the December-January-February (DJF) period is used. Both observed and forecast precipitation have a spatial resolution of 1°×1°.

[0106] like Figure 2 The following table shows the DJF forecast-observation correlation coefficient significance classification results. Grids with medium grayscale indicate a significantly positive correlation coefficient, indicating that the forecast has a certain indicative effect on observed precipitation. Grids with lighter grayscale indicate an insignificant correlation coefficient, while grids with darker grayscale indicate a significantly negative correlation coefficient between the forecast and observation. These two types of grids indicate suboptimal forecast performance. As can be seen, the majority of grids have a significantly negative correlation coefficient between the forecast and observation, followed by significantly positive correlation coefficients.

[0107] like Figure 2 The figure below shows the significance classification results of the DJF teleconnection coefficient. Among them, the medium gray grid indicates that the correlation coefficient is significantly positive, the lighter gray grid indicates that the correlation coefficient is not significant, and the darker gray grid indicates that the correlation coefficient between the forecast and observation is significantly negative. Compared with the forecast-observation correlation coefficient, it can be seen that the significance classification results of the teleconnection coefficient are more continuously distributed around the world. Figure 2 and Figure 3 , we can see similarities in the distribution of significant coefficients in some regions. For example, in southern North America, northern South America, and eastern Africa, the forecast-observation correlation coefficient is significantly positive, and the teleconnection strength is relatively strong.

[0108] Figure 2 and Figure 3 The spatial distribution map shows the distribution of regions with better forecasting results, but it cannot reflect the degree of spatial aggregation of these grids. Figure 2 and Figure 3 The correspondence cannot further give the strength of the correspondence and the consistency of this correspondence in space.

[0109] exist Figure 2 Based on the above, we first calculate the spatial consistency probability of a significantly positive forecast-observation correlation coefficient. We classify the correlation coefficients of each grid and all grids within a 5° radius around it, and then calculate the spatial consistency probability of a significantly positive grid based on the spatial weight coefficient matrix.

[0110] Figure 4This is the spatial consistency probability distribution map for significantly positive forecast-observation correlation coefficients. It can be seen that significantly positive correlation coefficients are more likely in southern North America, northern and southeastern South America, eastern and southern Africa, northeastern Asia, southern my country, Southeast Asia, southern Australia, and Europe. This indicates that precipitation forecasts are effective in these regions.

[0111] Further, establish Figure 2 and Figure 3 Specifically, Figure 5 The figure shows the spatial consistency probability distribution of significantly positive forecast-observation correlation coefficients and teleconnection coefficients for each grid. As can be seen, the probability is higher in southern North America, southeastern South America, eastern Africa, central Asia, and southern my country. This result reflects a strong correlation between forecast precipitation and teleconnection strength in these regions. Figure 6 This gives the probability of spatial consistency when the forecast-observation correlation coefficient is significantly positive but the teleconnection coefficient is insignificant. This probability is higher in northern Eurasia, southern Australia, and northwestern Africa, reflecting that forecasts can still be effective in areas with weak teleconnections. Figure 7 The probability of spatial consistency of a significantly positive forecast-observation correlation coefficient and a significantly negative teleconnection coefficient is given. The probability values ​​are higher in northeastern South America, southern Africa, Southeast Asia, and northeastern Asia.

[0112] Figure 5 、 Figure 6 and Figure 7 The spatial consistency probability of Figure 4 The spatial consistency probability shown is decomposed into three parts, which facilitates the analysis of different possible factors affecting the forecast effect. The above experimental results show that the spatial probability analysis method proposed in this invention can effectively quantify the spatial consistency probability of a significantly positive forecast-observation correlation coefficient, and at the same time decompose it into spatial consistency probabilities corresponding to different teleconnection relationships, which can intuitively display the spatial distribution of different corresponding relationships and provide a reference for the business use of forecasts.

[0113] Example 3

[0114] This embodiment proposes a spatial probability analysis system based on the corresponding relationship between precipitation forecast and teleconnection, and applies the spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection proposed in Example 1.

[0115] like Figure 8 , which is an architecture diagram of the spatial probability analysis system based on the corresponding relationship between precipitation forecast and teleconnection in this embodiment.

[0116] The spatial probability analysis system based on the corresponding relationship between precipitation forecast and teleconnection proposed in this embodiment includes:

[0117] The data acquisition module 1 is used to obtain the sample sequence of precipitation forecast to be analyzed, and the corresponding sample sequence of observed precipitation and meteorological factors.

[0118] The correlation coefficient calculation module 2 is used to calculate the forecast-observation correlation coefficient and the meteorological factor-observation precipitation telecorrelation coefficient of each grid in the target area based on the acquired sample sequence.

[0119] Classification module 3 is used to analyze the significance of the forecast-observation correlation coefficient and the meteorological factor-observation precipitation telecorrelation coefficient, and classify each grid according to the analysis results.

[0120] The significance judgment module 4 is used to judge the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification result.

[0121] The spatial weight coefficient calculation module 5 is used to calculate the spatial weight coefficient according to the spatial coordinates of the grid to obtain a spatial weight coefficient matrix.

[0122] The spatial consistency probability analysis module 6 is used to calculate the spatial consistency probability that the forecast-observation correlation coefficient is significantly positively correlated, and the spatial consistency probability of the corresponding relationship between the forecast-observation correlation coefficient and different remote connection coefficients based on the spatial weight coefficient matrix and the corresponding relationship between the forecast-observation correlation coefficient and the remote connection coefficient.

[0123] In an optional embodiment, the classification module 3 performs significance judgment and classification on each grid in the target area based on a preset significance level α and the forecast-observation correlation coefficient and meteorological factor-observed precipitation telecorrelation coefficient of each grid:

[0124] For the forecast-observation correlation coefficient r(o,f):

[0125] If the forecast-observation correlation coefficient r(o,f) is greater than r 1-α / 2 , it is judged to be significantly positively correlated;

[0126] If the forecast-observation correlation coefficient r(o,f) is less than or equal to r 1-α / 2 , and greater than r α / 2 , it is judged as not significant;

[0127] If the forecast-observation correlation coefficient r(o,f) is less than r α / 2 , it is judged to be significantly negatively correlated;

[0128] For meteorological factor-observation precipitation teleconnection coefficient r(o,η):

[0129] If the teleconnection coefficient r(o,η) is greater than r 1-α / 2 , it is judged to be significantly positively correlated;

[0130] If the teleconnection coefficient r(o,η) is less than or equal to r 1-α / 2 , and greater than r α / 2 , it is judged as not significant;

[0131] If the teleconnection coefficient r(o,η) is less than r α / 2 , it is judged to be a significant negative correlation.

[0132] In an optional embodiment, the significance judgment module 4 determines, grid by grid, for the case where the forecast-observation correlation coefficient r(o,f) is significantly positively correlated, whether the forecast-observation correlation is significantly positively correlated and the meteorological factor-observed precipitation telecorrelation is significantly positively correlated, the forecast-observation correlation is significantly positively correlated and the meteorological factor-observed precipitation telecorrelation is not significant, or the forecast-observation correlation is significantly positively correlated and the meteorological factor-observed precipitation telecorrelation is significantly negatively correlated, and constructs a corresponding relationship vector through Boolean numbers.

[0133] In an optional embodiment, the spatial weight coefficient calculation module 5 further includes row normalization of each spatial weight coefficient, and the spatial weight coefficient matrix A.

[0134] In an optional embodiment, the spatial consistency probability analysis module 6 calculates the Boolean number vector of the forecast-observation correlation coefficient of each grid as a significantly positive correlation based on the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient of each grid in the target area, and then multiplies the Boolean number vector with the spatial weight coefficient corresponding to the grid to calculate the spatial consistency probability that the forecast-observation correlation coefficient of the corresponding grid is a significantly positive correlation.

[0135] Its expression is as follows:

[0136] P(P AC )=A·b(P AC )=[p i ] N×1

[0137] Where, b(P AC ) is a Boolean vector of grid forecast-observation correlation coefficients that are significantly positive, p i Indicates the spatial consistency probability that the forecast-observation correlation coefficient for grid i is significantly positive.

[0138] The forecast-observation correlation coefficient of each grid is a Boolean vector b(P AC ), including the Boolean vector b(P AC&P ENSO ), the Boolean vector b(P AC &ns ENSO ), a Boolean vector b(P AC &N ENSO ).

[0139] Obviously, the above embodiments of the present invention are merely examples for the purpose of clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the claims of the present invention.

Claims

1. A spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection, characterized in that: The following steps are involved: Obtain the sample sequence of precipitation forecasts to be analyzed, as well as the corresponding sample sequences of observed precipitation and meteorological factors; Based on the obtained sample sequence, the forecast-observation correlation coefficient and meteorological factor-observed precipitation teleconnection coefficient of each grid are calculated respectively, and each grid is classified according to the significance of the forecast-observation correlation coefficient and meteorological factor-observed precipitation teleconnection coefficient; Determine the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results; Calculate the spatial weight coefficient according to the spatial coordinates of the grid to obtain the spatial weight coefficient matrix; Based on the spatial weight coefficient matrix and the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient, the spatial consistency probability of the forecast-observation correlation coefficient being significantly positive was calculated; The steps for calculating the probability of spatial consistency of the forecast-observation correlation coefficient being significantly positive include: Based on the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient of each grid in the target area, the Boolean number vector of the forecast-observation correlation coefficient of each grid is calculated. Then, the Boolean number vector is multiplied by the spatial weight coefficient corresponding to the grid to calculate the spatial consistency probability that the forecast-observation correlation coefficient of the corresponding grid is significantly positively correlated. The expression is as follows: P(P AC )=A·b(P AC )=[p i ] N×1 Where, b(P AC ) is the grid forecast-observation correlation coefficient, which is a significant positive correlation P AC Boolean vector of p i represents the spatial consistency probability that the forecast-observation correlation coefficient of grid i is significantly positively correlated; A represents the spatial weight coefficient matrix.

2. The spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to claim 1 is characterized in that: The steps of calculating the forecast-observation correlation coefficient and meteorological factor-observation precipitation teleconnection coefficient of each grid respectively based on the obtained sample sequence include: Extract the forecast precipitation data and observed precipitation data of the target area grid based on the obtained sample sequence; The forecast-observation correlation coefficient r(o,f) is calculated grid by grid; its expression is as follows: In the formula, o k represents the observed precipitation data for the kth year, f k represents the forecast precipitation data for the kth year; represent the mean of historical observed precipitation data and the mean of historical forecast precipitation data respectively; The meteorological factor-observed precipitation teleconnection coefficient r(o,η) is calculated grid by grid; its expression is as follows: Where η k represents the meteorological factor index of the kth year, Represents the mean of historical meteorological factor index.

3. The spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to claim 1 is characterized in that: The steps for classifying each grid based on the significance of the forecast-observation correlation coefficient and the meteorological factor-observation precipitation teleconnection coefficient include: Based on the preset significance level α, as well as the forecast-observation correlation coefficient and meteorological factor-observation precipitation telecorrelation coefficient of each grid, significance judgment and classification are performed on each grid in the target area: For the forecast-observation correlation coefficient r(o,f): If the forecast-observation correlation coefficient r(o,f) is greater than r 1-α / 2 , it is judged to be significantly positively correlated; If the forecast-observation correlation coefficient r(o,f) is less than or equal to r 1-α / 2 , and greater than r α / 2 , it is judged as not significant; If the forecast-observation correlation coefficient r(o,f) is less than r α / 2 , it is judged to be significantly negatively correlated; For meteorological factor-observation precipitation teleconnection coefficient r(o,η): If the teleconnection coefficient r(o,η) is greater than r 1-α / 2 , it is judged to be significantly positively correlated; If the teleconnection coefficient r(o,η) is less than or equal to r 1-α / 2 , and greater than r α / 2 , it is judged as not significant; If the teleconnection coefficient r(o,η) is less than r α / 2 , it is judged to be significantly negatively correlated; Among them, r 1-α / 2 is the 100×(1-α / 2) quantile of the correlation coefficient r, r α / 2 is the 100×(α / 2) quantile of the correlation coefficient r.

4. The spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to claim 3 is characterized in that: The steps for determining the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results include: For the case where the forecast-observation correlation coefficient r(o,f) is significantly positive, grid by grid, we determine whether the forecast-observation correlation is significantly positive and the meteorological factor-observed precipitation telecorrelation is significantly positive, the forecast-observation correlation is significantly positive and the meteorological factor-observed precipitation telecorrelation is not significant, or the forecast-observation correlation is significantly positive and the meteorological factor-observed precipitation telecorrelation is significantly negative, and construct the corresponding relationship vector using Boolean numbers; its expression is as follows: b(P AC &P ENSO )=[x i ] N×1 b(P AC &ns ENSO )=[x i ] N×1 b(P AC &N ENSO )=[x i ] N×1 Where N is the total number of grids in the target area; b(P AC &P ENSO ) is a significant positive correlation between forecast and observation. AC The meteorological factor-observed precipitation telecorrelation is significantly positively correlated P ENSO Boolean vector, when the grid i satisfies P AC &P ENSO , then x i The value is 1, otherwise x i The value is 0; b(P AC &ns ENSO ) indicates that the forecast-observation correlation is a significant positive correlation P AC The meteorological factor-observed precipitation telecorrelation is not significant. ENSO Boolean vector, when the grid i satisfies P AC &ns ENSO , then x i The value is 1, otherwise x i The value is 0; b(P AC &N ENSO ) indicates that the forecast-observation correlation is a significant positive correlation P AC The meteorological factor-observed precipitation telecorrelation is significantly negatively correlated. ENSO Boolean vector, when the grid i satisfies P AC &N ENSO , then x i The value is 1, otherwise x i The value is 0.

5. The spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to claim 1 is characterized in that: The steps of calculating the spatial weight coefficient according to the spatial coordinates of the grid and obtaining the spatial weight coefficient matrix include: Mark the coordinates of the target area grid with the upper left corner of the target area as the origin; The spatial weight coefficient of any two grid coordinates is calculated using the quadratic attenuation function of the distance. The expression is as follows: Where, d ij is the grid point (u i ,v i ) and grid points (u j ,v j ), d is the bandwidth value of the weight coefficient; The spatial weight coefficient matrix is ​​constructed based on the spatial weight coefficients of any two grid coordinates, and its expression is as follows: In=[in ij ] N×N Where N is the total number of grids.

6. The spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to claim 5 is characterized in that: The following steps are also included: Perform row standardization on the spatial weight coefficient matrix A, and take row standardization for each spatial weight coefficient; Its expression is as follows:

7. The spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to claim 4 is characterized in that: The forecast-observation correlation coefficient of each grid is a Boolean vector b(P AC ), including the Boolean vector b(P AC &P ENSO ), the Boolean vector b(P AC &ns ENSO ), a Boolean vector b(P AC &N ENSO ).

8. The spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to claim 7 is characterized in that: The following steps are also included: According to the spatial weight coefficient matrix A and the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient, the spatial consistency probability of the forecast-observation correlation coefficient being insignificant and the forecast-observation correlation coefficient being significantly negatively correlated is calculated.

9. A spatial probability analysis system based on the corresponding relationship between precipitation forecast and teleconnection, applying the spatial probability analysis method based on the corresponding relationship between precipitation forecast and teleconnection according to any one of claims 1 to 8, characterized in that: include: The data acquisition module is used to obtain the sample sequence of precipitation forecasts to be analyzed, as well as the corresponding sample sequence of observed precipitation and meteorological factors; The correlation coefficient calculation module is used to calculate the forecast-observation correlation coefficient and meteorological factor-observation precipitation telecorrelation coefficient of each grid in the target area based on the acquired sample sequence; The classification module is used to analyze the significance of the forecast-observation correlation coefficient and the meteorological factor-observation precipitation telecorrelation coefficient, and classify each grid according to the analysis results; The significance judgment module is used to judge the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results; A spatial weight coefficient calculation module is used to calculate the spatial weight coefficient according to the spatial coordinates of the grid to obtain a spatial weight coefficient matrix; The spatial consistency probability analysis module is used to calculate the spatial consistency probability of the forecast-observation correlation coefficient being significantly positively correlated, as well as the spatial consistency probability of the corresponding relationship between the forecast-observation correlation coefficient and different teleconnection coefficients based on the spatial weight coefficient matrix and the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient.

Images