Western pacific ocean subtropical high pressure intensity sub-season prediction method and system based on similar sea temperature evolution background

Abstract

The invention relates to the field of climate prediction, and provides a western pacific ocean subtropical high pressure intensity sub-season prediction method and system based on a similar sea temperature evolution background, and the method comprises the steps: S1, collecting day-by-day data and week-by-week data of a western pacific ocean subtropical high pressure intensity index; s2, classifying evolution trend backgrounds of the sea temperature indexes every year by utilizing clustering analysis; s3-S4, for each sea temperature evolution trend background category, obtaining a corresponding cumulative probability distribution curve of the West Pacific Ocean subtropical high pressure intensity index of observation and mode return; s5, determining the background category of the real-time forecast sea temperature evolution trend; and S6, according to the determined real-time forecast sea temperature evolution trend background category, performing probability matching correction on the predicted western pacific ocean subtropical high pressure intensity index by using a corresponding observation and mode return cumulative probability distribution curve. The method can provide scientific and technological support for correctly mastering the high evolution trend of the XiTaisui and improving the accuracy rate of sub-seasonal climate prediction.

Description

Technical Field

[0001] This invention relates to the field of climate prediction, and in particular to a method and system for predicting the intensity of the Western Pacific subtropical high pressure based on similar sea surface temperature evolution backgrounds in the second season. Background Technology

[0002] The Western Pacific subtropical high (WSP) strongly influences the weather and climate changes in East Asia, the subtropical region, and adjacent areas due to its dominance over the movement of subtropical weather systems and the transport of water vapor. It is one of the major atmospheric circulation systems affecting China's weather and climate and has always been a focus of meteorological research and operational attention. The low-level southwesterly airflow located on the northwest side of the WSP transports a large amount of water vapor to East Asia, maintaining the water vapor supply to the main rain belt. The position, shape, and intensity of the WSP determine the large-scale quasi-static fronts in East Asia, and areas controlled by the WSP often experience persistent high temperatures.

[0003] The Western Pacific Subtropical High (WSP) exhibits significant seasonal and interannual variations. The Northwest Pacific is the region with the strongest interannual variability in the Northern Hemisphere's subtropical summer climate, and this variability of the WSP causes significant climate anomalies in East Asia. When the WSP ridge is positioned further south or its westernmost point is further west in summer, the East Asian summer monsoon circulation is weaker, resulting in above-average rainfall in the Yangtze-Huaihe River basin during the flood season. Conversely, when the ridge is positioned further north or its westernmost point is further east, the summer monsoon circulation is stronger, leading to below-average rainfall in the Yangtze-Huaihe River basin. Furthermore, the variability in the WSP's position and intensity is closely related to climate anomalies such as drought in areas like the Jiangnan region, low summer temperatures in Northeast China, and the formation and activity paths of typhoons.

[0004] On the other hand, the Western Pacific Subtropical High (WSP) not only exhibits significant seasonal to interannual scale variations but also marked subseasonal scale variability. Firstly, the most prominent climatic feature of WSP activity is its seasonal north-south advance and retreat. From winter to summer, the WSP experiences two northward jumps, closely related to the position of China's summer rain belt. Generally, the WSP moves northward in June, coinciding with the plum rain season in the Yangtze River basin; in July (sometimes August), the WSP moves further north, ending the plum rain season and bringing drought to areas south of the Yangtze River, with the main rain belt shifting northward to North China and Northeast China. Secondly, in summer, the WSP moves northward to the subtropical region, influenced by tropical convection and mid-to-high latitude circulation, resulting in intraseasonal oscillations of multiple frequencies. The subseasonal scale variability of the WSP, particularly its persistent anomalies, is closely related to the occurrence and maintenance of extreme weather and climate events in my country. For example, the sustained westward extension and stable maintenance of the WSP in the summer of 2006 led to severe high-temperature drought in Sichuan and Chongqing, and the anomalous westward extension and maintenance of the WSP during the low-temperature rain, snow, and freezing disaster in southern my country in early 2008. Furthermore, the differences in the intraseasonal variability of the Western Pacific subtropical high also have a significant impact on the seasonal drought and flood distribution patterns. For example, the abnormal summer precipitation distribution under the backgrounds of the two super El Niño events in 2016 and 1998 was influenced by factors such as the different intraseasonal variability of the Western Pacific subtropical high, thus showing obvious differences.

[0005] Given the significant impact of the Western Pacific Subtropical High (WSP) on weather and climate anomalies in my country, accurately grasping its intensity and location is crucial for improving the accuracy of short-term climate forecasts. Existing research indicates that the seasonal to interannual variability of the WSP has high predictability. The anomaly correlation coefficient for summer WSP intensity forecasts based on numerical models and physical statistical models can reach approximately 0.8. This predictability stems not only from the El Niño-Southern Oscillation (ENSO) cycle but also from local air-sea interactions in the western Pacific. Compared to the seasonal to interannual variability, the predictability and predictable signal sources of the subseasonal-scale anomalies of the WSP remain unclear.

[0006] With the continuous improvement of numerical model performance, climate system models have become the main tool for climate prediction. Assessments show that, compared to seasonal scales, numerical models have significantly lower predictive skill for the subseasonal variability of the Western Pacific Subtropical High (WSP) and exhibit obvious systematic biases. Most models predict an earlier northward jump in the WSP and a more northerly ridge position, directly leading to a weaker predicted intensity of the Meiyu season in the middle and lower reaches of the Yangtze River. Furthermore, while models have high accuracy in predicting the persistence of WSP anomalies, their ability to predict transitions is poor, making it difficult to meet operational needs. The weak predictive ability of the WSP, a key circulation system influencing weather and climate anomalies in my country, has become one of the main factors restricting the improvement of the accuracy of subseasonal climate predictions in my country. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method and system for predicting the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution backgrounds, in order to improve the problems of large deviations and relatively low prediction skills of the Western Pacific subtropical high in climate prediction.

[0008] The present invention adopts the following technical solution:

[0009] On the one hand, this invention provides a method for predicting the intensity of the western Pacific subtropical high-pressure system in the second season based on similar sea surface temperature evolution backgrounds, including:

[0010] S1. Collect daily and weekly data of the intensity index of the Western Pacific subtropical high, including both observational data and model report data.

[0011] S2. Collect observational and historical model return data of sea surface temperature index, and use cluster analysis to classify the evolution trend background of sea surface temperature index each year to obtain several sea surface temperature evolution trend background categories.

[0012] S3. For each sea surface temperature evolution trend background category obtained in step S2, using the observation data and model report data from step S1, obtain the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under the sea surface temperature evolution trend background category respectively; the first CDF cumulative probability distribution curve is the observed cumulative probability density distribution of the western Pacific subtropical high pressure intensity index, and the second CDF cumulative probability distribution curve is the model report cumulative probability density distribution of the western Pacific subtropical high pressure intensity index.

[0013] S4. Repeat step S3 to obtain the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under all sea surface temperature evolution trend background categories.

[0014] S5. When making real-time forecasts, the background category of the real-time forecast sea surface temperature evolution trend is determined based on the sea surface temperature index data of the previous several months.

[0015] S6. Based on the background category of the real-time forecast sea surface temperature evolution trend determined in step S5, the predicted Western Pacific subtropical high pressure intensity index is matched and corrected using the corresponding first CDF cumulative probability distribution curve and second CDF cumulative probability distribution curve obtained in steps S3 and S4, so as to obtain the forecast correction value of the Western Pacific subtropical high pressure intensity index.

[0016] In addition to any of the possible implementations described above, another implementation is provided in which, in step S1, the intensity index of the Western Pacific subtropical high is defined as the average value of the 850 hPa geopotential height field in the Western Pacific region, and the Western Pacific region ranges from 15°N to 25°N and from 115°E to 150°E.

[0017] In addition to any of the possible implementations described above, another implementation is provided in which, in step S2, the sea surface temperature index is adopted. The regional sea surface temperature index is based on the SEMAP2.0 operational system for monitoring, analyzing and forecasting ENSO, established by the National Climate Center of the China Meteorological Administration.

[0018] The method of classifying the annual sea surface temperature index evolution trend using cluster analysis is as follows:

[0019] The dataset has n samples and p-dimensional observations:

[0020]

[0021] Any two sample points x i ,x j The Euclidean distance between i,j = 1, 2, ..., n is denoted as d. ij =d(x i ,x j If n samples are divided into k clusters, then the two samples (indices i1 and i2) that are farthest apart from each other are selected as the initial cluster points.

[0022]

[0023] Then determine the next cluster point, index i3, such that the point with the smallest distance between i3 and i1, i2 is equal to the largest of all other points with smaller distances between i1, i2:

[0024]

[0025] By repeatedly performing the above process, as the number of iterations gradually increases, the clustering results gradually stabilize, thus determining all k initial cluster points and the classification C. k ,i=1,…,k.

[0026] In addition to any of the possible implementations described above, another implementation is provided where the number of clusters k = 5, i.e., 5 background categories of sea surface temperature evolution trends.

[0027] In addition to any of the possible implementations described above, another implementation is provided in which, in step S3, the method for determining the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under a certain sea surface temperature evolution trend background category is as follows:

[0028] The 850 hPa geopotential height field in the region (15°N-25°N, 115°E-150°E) is interpolated to the same horizontal resolution grid by using the bilinear interpolation algorithm, averaging the observations and model reports week by week. To ensure a sufficiently large statistical sample size, the horizontal resolution of the grid can be appropriately increased, provided that computational efficiency allows. Statistical analysis is performed on all grid points in the observations and model reports to obtain the first probability density distribution of the observed geopotential height values ​​at the grid points and the second probability density distribution of the geopotential height values ​​in the model reports at the grid points. This leads to the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve.

[0029] In addition to any of the possible implementations described above, another implementation is provided in which, in step S5, the real-time forecast is based on the forecast data for the five months preceding the model. The sea surface temperature index (SST) is used to determine the background classification of real-time forecast SST evolution trends. Specifically, the method involves using the Euclidean distance method to find the classification that most closely matches the SST evolution trend background of the five months prior to the model's reporting time. The Euclidean distance calculation formula is as follows:

[0030]

[0031] Where n=5, representing the sea surface temperature evolution trend 5 months in advance, y 1k and y 2k These are real-time forecasts that are 1-5 months ahead of time. The sea surface temperature index and K-means clustering yield the average value of sea surface temperature evolution trends; for a given cluster C k The Euclidean distance is the minimum value min(d). 12 When considering the real-time forecast of sea surface temperature evolution trends, it is believed that the cluster is most similar to this cluster.

[0032] In addition to any of the possible implementations described above, another implementation is provided in which the specific method for matching and correcting the predicted intensity index of the western Pacific subtropical high in step S6 is as follows:

[0033] The real-time forecast value of the Western Pacific subtropical high intensity index x at time d is... f (d) First, the corresponding second cumulative probability P is obtained from the second CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend. c (x f (d)); Then, find the first cumulative probability P on the first CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend. o (x)=P c (x f (d) The corresponding Western Pacific subtropical high intensity index x fcorr (d), the x fcorr(d) As the forecast correction value of the Western Pacific subtropical high intensity index; where the subscripts o, c, and f represent observational data, model report data, and real-time forecast data, respectively.

[0034] In addition to any of the possible implementations described above, a further implementation is provided, wherein the method further includes: S7, evaluating the forecast correction effect of the Western Pacific subtropical high pressure intensity index obtained in steps S1-S6.

[0035] On the other hand, the present invention also provides a subseasonal prediction system for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution backgrounds. This system is used to implement the above-mentioned method and includes:

[0036] The data collection unit is used to collect daily and weekly data of the Western Pacific subtropical high intensity index and historical model return data of the sea surface temperature index.

[0037] The sea surface temperature evolution trend background classification unit uses historical observation and model return data of the sea surface temperature index to classify the evolution trend background of the sea surface temperature index each year using cluster analysis.

[0038] The cumulative probability density distribution calculation unit uses daily and weekly data of the Western Pacific subtropical high pressure intensity index to calculate the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under each type of sea surface temperature evolution trend background category; the first CDF cumulative probability distribution curve is the observed cumulative probability density distribution of the Western Pacific subtropical high pressure intensity index, and the second CDF cumulative probability distribution curve is the model-reported cumulative probability density distribution of the Western Pacific subtropical high pressure intensity index.

[0039] The real-time forecast correction unit determines the background category of the real-time forecast sea surface temperature evolution trend based on historical data of the annual sea surface temperature index for several months prior to the forecast. It then uses the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend to match and correct the predicted intensity index of the western Pacific subtropical high.

[0040] In addition to any of the possible implementations described above, another implementation is provided in which the sea surface temperature index adopts... Sea surface temperature index.

[0041] The beneficial effects of this invention are as follows: The prediction method of this invention corrects numerical model predictions, effectively reducing the systematic bias of model predictions of the Western Pacific Subtropical High (WSP) intensity and improving the accuracy of sub-seasonal predictions. Currently, in operational monitoring and forecasting, the WSP is characterized by the 5800 gpm isopleths on the geopotential height field. Reducing the systematic bias of model predictions will enhance the comparability of observed and model-predicted WSP intensity, aiding in the analysis of atmospheric circulation trends. Furthermore, given the important role of the WSP in water vapor transport, precipitation location and intensity, and high-temperature drought, improvements in WSP sub-seasonal variability prediction techniques will provide effective support for improving the accuracy of sub-seasonal climate predictions. Attached Figure Description

[0042] Figure 1 The diagram shown is a flowchart of a method for predicting the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution backgrounds according to an embodiment of the present invention.

[0043] Figure 2 The diagram shown illustrates the principle of pattern error correction based on probability matching in this embodiment.

[0044] Figure 3 The flowchart shown is a pattern error correction method based on probability matching in the embodiment.

[0045] Figure 4 The figure shown is a diagram of the definition of the Western Pacific subtropical high intensity index and the CMA-CPSv3 model's subseasonal forecasting skill test in the embodiment; (a) definition of the Western Pacific subtropical high intensity index (gpm), (b) root mean square error (RMSE) and time correlation coefficient (TCC) of the Western Pacific subtropical high intensity predicted by the CMA-CPSv3 model for the next 4 weeks starting from June-July, and (c) probability density distribution of the Western Pacific subtropical high intensity for the next 4 weeks as observed (red) and predicted by the model (blue).

[0046] Figure 5 The figure shows the relationship between the prediction bias (vertical axis) and intensity (horizontal axis) of the Western Pacific subtropical high predicted by the CMA-CPSv3 model from June to July for the next four weeks.

[0047] Figure 6 The figure shows the TCC distribution of the 850hPa geopotential height field predicted by CMA-CPSv3 from May to September in the example (the black dotted area is the area that passed the 95% confidence test and the prediction lead time is 0 months).

[0048] Figure 7 The figure shows the background classification of the sea surface temperature index evolution trend in the Nino3.4 region based on the K-means clustering method in the embodiment.

[0049] Figure 8The figure shows the weekly probability density distribution of the Western Pacific subtropical high intensity in June and July during the summer months under different ENSO evolution trends (C1 to C5) in the examples.

[0050] Figure 9 The figure shows a comparison of techniques before and after the CMA-CPSv3 model's prediction of the intensity of the Western Pacific subtropical high for the next four weeks starting in June-July in the example; the Western Pacific subtropical high intensity bias (red) and probability mapping correction (green) and similar sea surface temperature background correction (blue); the left figure is the root mean square error of the prediction, and the right figure is the time correlation coefficient. Detailed Implementation

[0051] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that the technical features or combinations of technical features described in the following embodiments should not be considered in isolation, but can be combined with each other to achieve better technical effects.

[0052] like Figure 1 As shown, an embodiment of the present invention provides a method for predicting the intensity of the western Pacific subtropical high based on similar sea surface temperature evolution backgrounds during the second season, comprising:

[0053] S1. Collect daily and weekly data of the intensity index of the Western Pacific subtropical high, including both observational data and model report data.

[0054] The daily and weekly data were obtained from the China Meteorological Administration's CRA reanalysis and model return 850hPa geopotential height field data.

[0055] S2. Collect historical observation and model return data of sea surface temperature index, and use cluster analysis to classify the evolution trend background of sea surface temperature index each year to obtain several sea surface temperature evolution trend background categories.

[0056] S3. For each sea surface temperature evolution trend background category obtained in step S2, using the observation data and model report data from step S1, obtain the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under the sea surface temperature evolution trend background category respectively; the first CDF cumulative probability distribution curve is the observed cumulative probability density distribution of the western Pacific subtropical high pressure intensity index, and the second CDF cumulative probability distribution curve is the model report cumulative probability density distribution of the western Pacific subtropical high pressure intensity index.

[0057] S4. Repeat step S3 to obtain the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under all sea surface temperature evolution trend background categories.

[0058] S5. When making real-time forecasts, the background category of the real-time forecast sea surface temperature evolution trend is determined based on the sea surface temperature index data of the previous several months.

[0059] S6. Based on the background category of the real-time forecast sea surface temperature evolution trend determined in step S5, the predicted Western Pacific subtropical high pressure intensity index is matched and corrected using the corresponding first CDF cumulative probability distribution curve and second CDF cumulative probability distribution curve obtained in steps S3 and S4, so as to obtain the forecast correction value of the Western Pacific subtropical high pressure intensity index.

[0060] In one specific embodiment, in step S1, the intensity index of the Western Pacific subtropical high pressure is defined as the average value of the 850 hPa geopotential height field in the Western Pacific region, which is located in the range of 15°N-25°N and 115°E-150°E.

[0061] In one specific embodiment, in step S2, the sea surface temperature index is adopted. The regional sea surface temperature index is based on the SEMAP2.0 operational system for monitoring, analyzing and forecasting ENSO, established by the National Climate Center of the China Meteorological Administration.

[0062] The method of classifying the annual sea surface temperature index evolution trend using cluster analysis is as follows:

[0063] The dataset has n samples and p-dimensional observations:

[0064]

[0065] Any two sample points x i ,x j The Euclidean distance between i,j = 1, 2, ..., n is denoted as d. ij =d(x i ,x j If n samples are divided into k clusters, then the two samples (indices i1 and i2) that are farthest apart from each other are selected as the initial cluster points.

[0066]

[0067] Then determine the next cluster point, index i3, such that the point with the smallest distance between i3 and i1, i2 is equal to the largest of all other points with smaller distances between i1, i2:

[0068]

[0069] By repeatedly performing the above process, as the number of iterations gradually increases, the clustering results gradually stabilize, thus determining all k initial cluster points and the classification C. k ,i=1,…,k.

[0070] Specifically, in this application, x iLet x be the sea surface temperature index sample point for year i. ip C represents the sea surface temperature index for month p of year i (the month p before the reporting date). k For the year sample points of the k-th cluster.

[0071] In one specific embodiment, the number of clusters k = 5, that is, 5 background categories of sea surface temperature evolution trends.

[0072] In a specific embodiment, in step S3, the method for determining the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under a certain sea surface temperature evolution trend background category is as follows:

[0073] The 850 hPa geopotential height field in the (15°N-25°N, 115°E-150°E) region, calculated weekly averages from observations and model reports, is interpolated to a grid with the same horizontal resolution. To ensure a sufficiently large statistical sample size, the horizontal resolution of the grid can be appropriately increased, provided computational efficiency allows. Statistics are performed on all grid points from both observations and model reports to obtain the first probability density distribution of the observed geopotential height values ​​and the second probability density distribution of the model-reported geopotential height values. This leads to the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve, as shown below. Figure 3 As shown.

[0074] Since the probability density distribution of geopotential height approximates a normal distribution, the probability density distribution (PDF) of the model and observation data is converted into a cumulative probability distribution (CDF) in the correction process. This makes the distribution function form monotonic and also ensures the uniqueness of the mapping result.

[0075] In one specific embodiment, in step S5, the real-time forecast is based on the model for the five months preceding the forecast. The sea surface temperature index (SST) is used to determine the background classification of real-time forecast SST evolution trends. Specifically, the method involves using the Euclidean distance method to find the classification that most closely matches the SST evolution trend background of the five months prior to the model's reporting time. The Euclidean distance calculation formula is as follows:

[0076]

[0077] Where n=5, representing the sea surface temperature evolution trend 5 months in advance, y 1k and y 2k These are real-time forecasts that are 1-5 months ahead of time. The sea surface temperature index and K-means clustering yield the average value of sea surface temperature evolution trends; for a given cluster C k The Euclidean distance is the minimum value min(d). 12 When considering the real-time forecast of sea surface temperature evolution trends, it is believed that the cluster is most similar to this cluster.

[0078] In one specific embodiment, such as Figure 2 , Figure 3 As shown, the specific method for matching and correcting the predicted intensity index of the western Pacific subtropical high in step S6 is as follows:

[0079] The real-time forecast value of the Western Pacific subtropical high intensity index x at time d is... f (d) First, the corresponding second cumulative probability P is obtained from the second CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend. c (x f (d)); Then, find the first cumulative probability P on the first CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend. o (x)=P c (x f (d) corresponds to the intensity index of the Western Pacific subtropical high pressure x fcorr (d), the x fcorr (d) As the forecast correction value of the Western Pacific subtropical high intensity index; where the subscripts o, c, and f represent observational data, model report data, and real-time forecast data, respectively.

[0080] In one specific embodiment, the method further includes: S7, evaluating the forecast correction effect of the Western Pacific subtropical high pressure intensity index obtained in steps S1-S6.

[0081] On the other hand, the present invention also provides a subseasonal prediction system for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution backgrounds. This system is used to implement the above-mentioned method and includes:

[0082] The data collection unit is used to collect daily and weekly data of the Western Pacific subtropical high intensity index and sea surface temperature index data from observations and model reports.

[0083] The sea surface temperature evolution trend background classification unit uses historical observation data of the sea surface temperature index to classify the evolution trend background of the sea surface temperature index each year using cluster analysis.

[0084] The cumulative probability density distribution calculation unit uses daily and weekly data of the Western Pacific subtropical high pressure intensity index to calculate the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under each type of sea surface temperature evolution trend background category; the first CDF cumulative probability distribution curve is the observed cumulative probability density distribution of the Western Pacific subtropical high pressure intensity index, and the second CDF cumulative probability distribution curve is the model-reported cumulative probability density distribution of the Western Pacific subtropical high pressure intensity index.

[0085] The real-time forecast correction unit determines the background category of the real-time forecast sea surface temperature evolution trend based on historical data of the annual sea surface temperature index for several months prior to the forecast. It then uses the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend to match and correct the predicted intensity index of the western Pacific subtropical high.

[0086] In one specific embodiment, the sea surface temperature index is adopted. Sea surface temperature index.

[0087] Since December 2024, the China Meteorological Administration's integrated climate model prediction system (CMA-CPSv3) at sub-seasonal-seasonal-interannual scales has been officially operational, generating forecast graphic products to meet national and provincial operational service needs. The time scale includes monthly forecasts for the next seven months at different reporting start dates, and the spatial scale includes global, Asian, and Chinese scales. Elements include key elements of the atmosphere, land surface, ocean, and sea ice sphere, as well as important climate indices. Since its pre-operational launch in March 2021, the system has completed historical backcalculation experiments covering 365 days in the sub-seasonal prediction subsystem (15 years) and 12 reporting months in the seasonal prediction subsystem (20 years). Data shows that the system's forecasting capabilities for key seasonal indicators such as temperature, precipitation, El Niño-Southern Oscillation (ENSO) index, Asian summer monsoon index, and Western Pacific subtropical high index in my country are superior to the previous version of the seasonal prediction subsystem. Its ability to predict temperature and precipitation at phenological and monthly scales, as well as its forecasting skills for the tropical atmospheric intraseasonal oscillation (MJO), are generally superior to the existing operational system of the lunar dynamic extension forecasting model (DERF2.0), but its performance in predicting the subseasonal variability of the Western Pacific subtropical high is still unclear.

[0088] Atmospheric circulation observation data are from the first-generation global atmosphere / land surface reanalysis (CRA-40) data released by the National Meteorological Information Center of the China Meteorological Administration. The data length is since 1979, with a daily temporal resolution, a spatial resolution of 34 km, and 64 vertical layers. Model data uses the CMA-CPSv3 model's sub-seasonal historical return data, covering the period from 2006 to 2024, selecting data from Mondays and Thursdays in June and July each year, with a forecast duration of 60 days.

[0089] Because the Western Pacific subtropical high is a deep anticyclonic circulation system in the lower troposphere over the northwestern Pacific, it can be observed in the standard deviation field of the summer mean 850 hPa geopotential height. Figure 4 (a) A large value center exists in the mid-latitudes of the Northwest Pacific. Based on the definitions in relevant literature, this application uses the average value of the Northwest Pacific (20°N-30°N, 115°N-150°N) in the 850hPa geopotential height field as the intensity index of the Western Pacific subtropical high, and obtains the weekly average result based on daily data.

[0090] First, the predictive skill and error distribution of the CMA-CPSv3 model for the subseasonal variability of the Western Pacific subtropical high (SPH) intensity are analyzed. The evaluation results show that the model's weekly predictive skill for the SPH decreases over time, with the time correlation coefficient (TCC) at 0.36 in the first week and falling below 0.1 in the fourth week. Simultaneously, the model's prediction bias continues to increase over time, with the root mean square error (RMSE) of the predicted intensity compared to the observed intensity at 130 gpm in the first week, exceeding 160 gpm in the third and fourth weeks. Figure 4 (b) From the probability density distributions (PDFs) of the model-reported and actual Western Pacific subtropical high intensity, it can be seen that the model-reported Western Pacific subtropical high intensity is weaker than the actual system, and the variance decreases continuously with the increase of the forecast lead time. Figure 4 (cf). On the other hand, the model's predicted intensity bias of the Western Pacific subtropical high is directly proportional to its intensity, and this correlation strengthens with increasing forecast lead time; the model's predicted bias of the Western Pacific subtropical high is inversely proportional to the actual intensity, meaning the stronger the actual Western Pacific subtropical high, the weaker the model's predicted Western Pacific subtropical high. Figure 5 ).

[0091] Existing research indicates that ENSO typically influences the Western Pacific subtropical high (WPS) via the Northwest Pacific anticyclone (cyclone), thereby affecting precipitation anomalies in the East Asian monsoon region. The CMA-CPSv3 model achieves a TCC exceeding 0.6 for the monthly 850 hPa geopotential height field over the Northwest Pacific from May to September, passing a 95% confidence level test, indicating high forecasting skill for the monthly scale variability of the WPS intensity. Figure 6 Since the predictable signal sources of the Western Pacific Subtropical High mainly originate from ENSO, the intensity probability distribution of the Western Pacific Subtropical High varies under different ENSO phases and evolution trends. This invention aims to correct the error of the sub-seasonal variability of the Western Pacific Subtropical High predicted by the model under tropical sea surface temperature background.

[0092] For forecast years that are 0-5 months ahead The index evolution trend was analyzed using K-means clustering to categorize historical reporting years into 5 classes. Figure 7 Based on the classification, the probability density distribution of the intensity of the Western Pacific subtropical high was calculated under different ENSO evolution trends. It can be seen that when... The index shows a downward trend, meaning that during the El Niño decay phase, the intensity of the Western Pacific subtropical high is systematically stronger. Figure 8 ).

[0093] To verify the effectiveness of probability mapping bias correction schemes that consider different ENSO evolution trends, two sets of correction schemes were designed for comparative analysis. The first set calculates the weekly probability density distribution of the Western Pacific subtropical high without considering the ENSO sea surface temperature background when calculating the observed and model-predicted probability density distributions; the second set calculates the probability density distribution based on the similarity of ENSO evolution trends obtained through clustering. Figure 9 The weekly forecast bias of the Western Pacific subtropical high intensity based on the original output of the CMA-CPSv3 model, the bias after probability mapping correction, and the correction results considering the trend similarity of ENSO evolution in previous years are presented. It can be seen that, compared to the original output, the probability mapping correction method can effectively reduce the model's sub-seasonal forecast bias, and by taking into account the ENSO evolution trend, the root mean square error between the forecast and the actual situation can be further reduced.

[0094] This invention utilizes the observational fact that the seasonal variation in the intensity of the Western Pacific Subtropical High (WSP) is closely related to anomalies in tropical air-sea systems such as El Niño and the Southern Oscillation (ENSO). Addressing the spatiotemporal differences in the subseasonal prediction bias of the WSP under different ENSO evolution trends in numerical models, a probabilistic mapping method is employed for correction, effectively reducing prediction bias and improving prediction accuracy. Given the significant impact of the WSP on weather and climate anomalies in my country, this invention provides scientific and technological support for accurately understanding the evolution trend of the WSP and improving the accuracy of subseasonal climate prediction.

[0095] While several embodiments of the present invention have been provided herein, those skilled in the art should understand that modifications can be made to these embodiments without departing from the spirit of the invention. The above embodiments are merely exemplary and should not be construed as limiting the scope of the invention.

Claims

1. A method for predicting the intensity of the western Pacific subtropical high-pressure system in a subseasonal manner based on similar sea surface temperature evolution backgrounds, the method comprising: S1. Collect daily and weekly data of the Western Pacific subtropical high intensity index, including both observational data and model report data. S2. Collect historical observation and model return data of sea surface temperature index, and use cluster analysis to classify the evolution trend background of sea surface temperature index each year to obtain several sea surface temperature evolution trend background categories. S3. For each sea surface temperature evolution trend background category obtained in step S2, the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under the sea surface temperature evolution trend background category are calculated using the observation data and model report data from step S1. The first CDF cumulative probability distribution curve is the observed cumulative probability distribution of the western Pacific subtropical high pressure intensity index, and the second CDF cumulative probability distribution curve is the model report of the western Pacific subtropical high pressure intensity index. S4. Repeat step S3 to obtain the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under all sea surface temperature evolution trend background categories. S5. When making real-time forecasts, the background category of the real-time forecast sea surface temperature evolution trend is determined based on the sea surface temperature index data of the previous several months. S6. Based on the background category of the real-time forecast sea surface temperature evolution trend determined in step S5, the predicted Western Pacific subtropical high pressure intensity index is matched and corrected using the corresponding first CDF cumulative probability distribution curve and second CDF cumulative probability distribution curve obtained in steps S3 and S4, so as to obtain the forecast correction value of the Western Pacific subtropical high pressure intensity index.

2. The method for predicting the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution background as described in claim 1, characterized in that, In step S1, the intensity index of the Western Pacific subtropical high pressure is defined as the average value of the 850 hPa geopotential height field in the Western Pacific region, which is located in the range of 15°N-25°N and 115°E-150°E.

3. The subseasonal prediction method for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution background as described in claim 1, characterized in that, In step S2, the sea surface temperature index is adopted. Sea surface temperature index in zone 3.4, data source: SEMAP2.0, an operational system for monitoring, analyzing and forecasting ENSO established by the National Climate Center of the China Meteorological Administration; The method of classifying the annual sea surface temperature index evolution trend using cluster analysis is as follows: The dataset has n samples and p-dimensional observations: Any two sample points x i ,x j The Euclidean distance between i,j = 1, 2, ..., n is denoted as d. ij =d(x i ,x j If n samples are divided into k clusters, then the two sample points with the greatest distance between all sample points are selected as the initial cluster points, with the indexes i1 and i2. Then determine the next cluster point, index i3, such that the point with the smallest distance between i3 and i1, i2 is equal to the largest of all other points with smaller distances between i1, i2: By repeatedly performing the above process, as the number of iterations gradually increases, the clustering results gradually stabilize, thus determining all k initial cluster points and the classification C. k ,i=1,…,k.

4. The subseasonal prediction method for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution background as described in claim 3, characterized in that, The number of clusters k = 5, which means there are 5 background categories for sea surface temperature evolution trends.

5. The subseasonal prediction method for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution background as described in claim 1, characterized in that, In step S3, the method for determining the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under a certain sea surface temperature evolution trend background category is as follows: The 850 hPa geopotential height field in the region (15°N-25°N, 115°E-150°E), which is averaged weekly by week from observations and model reports, is interpolated to the same horizontal resolution grid. Statistical analysis is performed on all grid points from observations and model reports to obtain the first probability density distribution of the observed geopotential height values ​​at the grid points and the second probability density distribution of the geopotential height values ​​from model reports at the grid points. This leads to the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve.

6. The subseasonal prediction method for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution background as described in claim 1, characterized in that, In step S5, during real-time forecasting, the forecast for the five months preceding the start of the model is based on... 3.4 Sea surface temperature index (SST) is used to determine the background classification of real-time forecast SST evolution trends. The specific method is as follows: Based on the Euclidean distance method, the classification that most closely matches the SST evolution trend background of the five months prior to the model's reporting time is identified. The Euclidean distance calculation formula is as follows: Where n=5, representing the sea surface temperature evolution trend 5 months in advance, y 1k and y 2k These are real-time forecasts that are 1-5 months ahead of time. 3.4 The sea surface temperature index of the region and the K-means clustering are used to obtain the average value of the sea surface temperature evolution trend; for a certain cluster C k The Euclidean distance is the minimum value min(d). 12 When considering the real-time forecast of sea surface temperature evolution, it is believed that the background of this cluster is most similar to that of the sea surface temperature evolution trend.

7. The subseasonal prediction method for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution background as described in claim 1, characterized in that, In step S6, the specific method for matching and correcting the predicted intensity index of the western Pacific subtropical high is as follows: The real-time forecast value of the Western Pacific subtropical high intensity index x at time d is... f (d) First, the corresponding second cumulative probability P is obtained on the second CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend. c (x f (d)); Then, on the first CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend, find the first cumulative probability P. o (x)=P c (x f (d) The corresponding Western Pacific subtropical high intensity index x fcorr (d), the x fcorr (d) As the forecast correction value of the Western Pacific subtropical high intensity index; where the subscripts o, c, and f represent observational data, model report data, and real-time forecast data, respectively.

8. The subseasonal prediction method for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution background as described in claim 1, characterized in that, The method further includes: S7, evaluating the forecast correction effect of the Western Pacific subtropical high pressure intensity index in steps S1-S6.

9. A subseasonal prediction system for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution backgrounds, characterized in that, The system is used to implement the method as described in any one of claims 1-8, the system comprising: The data collection unit is used to collect daily and weekly data of the Western Pacific subtropical high intensity index and sea surface temperature index data from observations and model reports. The sea surface temperature evolution trend background classification unit uses historical observation data of the sea surface temperature index to classify the evolution trend background of the sea surface temperature index each year using cluster analysis. The cumulative probability density distribution calculation unit uses daily and weekly data of the Western Pacific subtropical high pressure intensity index to calculate the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve under each type of sea surface temperature evolution trend background category; the first CDF cumulative probability distribution curve is the observed cumulative probability density distribution of the Western Pacific subtropical high pressure intensity index, and the second CDF cumulative probability distribution curve is the model-reported cumulative probability density distribution of the Western Pacific subtropical high pressure intensity index. The real-time forecast correction unit determines the background category of the real-time forecast sea surface temperature evolution trend based on historical data of the annual sea surface temperature index for several months prior to the forecast. It then uses the first CDF cumulative probability distribution curve and the second CDF cumulative probability distribution curve corresponding to the background category of the real-time forecast sea surface temperature evolution trend to match and correct the predicted intensity index of the western Pacific subtropical high.

10. The subseasonal prediction system for the intensity of the Western Pacific subtropical high based on similar sea surface temperature evolution background as described in claim 9, characterized in that, The sea surface temperature index is adopted. Sea surface temperature index in zone 3.4.

Images