Identification and non-stationary characteristics analysis methods of regional heavy rainfall events caused by tropical cyclones

Abstract

The present invention discloses a method for identifying and analyzing non-steady-state characteristics of regional heavy rainfall events caused by tropical cyclones. The method comprises: data collection; identification of regional heavy rainfall events; correlation between tropical cyclones and regional heavy rainfall events; identification of characteristics of regional heavy rainfall caused by tropical cyclones; variability analysis of regional heavy rainfall characteristics caused by tropical cyclones; non-steady-state frequency analysis of regional heavy rainfall characteristics caused by tropical cyclones; mechanism and impact of shortening the recurrence period of extreme regional heavy rainfall events caused by tropical cyclones. The present invention has the beneficial effects of fully identifying the spatiotemporal evolution and movement characteristics of regional extreme rainfall caused by tropical cyclones, clarifying the changing laws of the recurrence period of extreme regional heavy rainfall events caused by tropical cyclones, helping to further improve the accuracy of tropical cyclone-induced heavy rainfall forecasts, and providing a scientific basis for resisting disasters caused by tropical cyclones and making decisions.

Description

Technical Field

[0001] The present invention relates to a method for identifying heavy rainfall events in a tropical cyclone-induced region and analyzing non-steady-state characteristics. Background Art

[0002] Heavy precipitation events, defined as precipitation exceeding a specified threshold, have garnered widespread attention due to their significant impact on society and the economy, and are a key research topic in the Earth sciences. Regional heavy precipitation events are defined as heavy precipitation occurring simultaneously over a large spatial area. In a warming climate, tropical cyclones tend to become stronger, decay more slowly, and last longer, increasing precipitation variability in monsoon regions, leading to greater precipitation intensity and a wider range. This increases the incidence of regional heavy precipitation events, increasing the risk of triggering large-scale severe floods, and posing a significant threat to regional agriculture, the socio-economic system, and the safety of people and property.

[0003] Tropical cyclones are a significant source of heavy rainfall, contributing over 30% of heavy rainfall in coastal areas worldwide and over 50% of heavy rainfall in China. However, current research on the identification of heavy rainfall events is based on single-point data (i.e., stations or grid points), ignoring the spatial connectivity and temporal continuity of heavy rainfall events, and lacking an exploration of the role of tropical cyclones in regional heavy rainfall events. Therefore, it is necessary to accurately identify the movement and evolution of regional heavy rainfall caused by tropical cyclones. This will further improve the accuracy of tropical cyclone-induced heavy rainfall forecasts and provide a scientific basis for disaster prevention and decision-making related to tropical cyclone-induced disasters. Summary of the Invention

[0004] The purpose of the present invention is to propose a method for identifying and analyzing the non-steady-state characteristics of regional heavy rainfall events caused by tropical cyclones, so as to solve the technical problem of insufficient research methods on the impact of tropical cyclones on regional heavy rainfall events.

[0005] The present invention provides a method for identifying heavy rainfall events in a tropical cyclone-induced region and analyzing non-steady-state characteristics, comprising the following steps:

[0006] S1. Data collection: Collect observation-based gridded precipitation data, tropical cyclone best track data, and meteorological reanalysis data;

[0007] S2. Identification of regional heavy rainfall events: Based on the definition of heavy rainfall and the method for determining regional heavy rainfall events, regional heavy rainfall events are identified in combination with the grid precipitation data collected in step S1;

[0008] S3. Correlation between tropical cyclones and regional heavy rainfall events: Based on the regional heavy rainfall events obtained in step S2, a multi-temporal clustering algorithm is used to obtain a mapping relationship between tropical cyclones and the regional heavy rainfall events they trigger.

[0009] S4. Identification of characteristics of regional heavy rainfall caused by tropical cyclones: Based on the mapping relationship between tropical cyclones and regional heavy rainfall events caused by them obtained in step S3, the spatiotemporal evolution of the characteristics of regional heavy rainfall events caused by tropical cyclones is quantified from four aspects: duration, cumulative affected area, cumulative average precipitation depth, and cumulative total precipitation.

[0010] S5. Analysis of the variability of regional heavy rainfall characteristics caused by tropical cyclones: combining the characteristics of regional heavy rainfall events caused by tropical cyclones obtained in step S4, quantifying the variability of event characteristics, and analyzing the mechanism affecting the characteristic variability;

[0011] S6. Non-steady-state frequency analysis of regional heavy rainfall characteristics caused by tropical cyclones: Based on the regional heavy rainfall events caused by tropical cyclones obtained in step S3, extreme value theory is used to estimate the recurrence period of the annual maximum cumulative impact area, cumulative area average precipitation depth, and cumulative total precipitation of regional heavy rainfall events in step S4, to test whether the probability of occurrence of extreme regional heavy rainfall events caused by tropical cyclones has changed;

[0012] S7. Mechanism and impact of shortened return periods of extreme regional heavy rainfall events caused by tropical cyclones: Based on the changes in the characteristic return periods of regional heavy rainfall events caused by tropical cyclones obtained in step S6, analyze the factors affecting the shortened return period.

[0013] A storage medium stores instructions and data for implementing a method for identifying heavy rainfall events in a tropical cyclone-induced region and analyzing non-steady-state characteristics.

[0014] A device for identifying heavy rainfall events in a tropical cyclone-induced region and analyzing non-steady-state characteristics comprises: a processor and a storage medium; the processor loads and executes instructions and data in the storage medium to implement a method for identifying heavy rainfall events in a tropical cyclone-induced region and analyzing non-steady-state characteristics.

[0015] The beneficial effects provided by the present invention are:

[0016] (1) The present invention discloses a method for identifying regional heavy rainfall events caused by tropical cyclones, determines the mapping relationship between tropical cyclones and regional heavy rainfall events caused by tropical cyclones, and quantifies the spatiotemporal evolution characteristics of this type of heavy rainfall events in the historical period from four aspects: duration, cumulative affected area, cumulative average precipitation depth and cumulative total precipitation. The method of the present invention can fully identify the spatiotemporal evolution and movement characteristics of regional extreme precipitation caused by tropical cyclones, which will further improve the accuracy of heavy rainfall forecasts caused by tropical cyclones.

[0017] (2) The non-steady-state characteristics of regional heavy rainfall events caused by tropical cyclones were analyzed. The non-steady-state frequency of extreme regional heavy rainfall events caused by tropical cyclones was quantified from four aspects: duration, cumulative affected area, cumulative average precipitation depth and cumulative total precipitation. The extreme value theory was used to predict the future recurrence period of extreme regional heavy rainfall events caused by tropical cyclones, and the mechanism and impact of the shortening of the recurrence period were explored, providing a scientific basis for resisting disasters caused by tropical cyclones and making decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic flow chart of the method of the present invention;

[0019] Figure 2 It is a framework for identifying regional heavy rainfall events caused by tropical cyclones;

[0020] Figure 3 This is the distribution map of regional heavy rainfall events in the eastern monsoon region of China;

[0021] Figure 4 This is a skewed distribution map of the characteristics of regional heavy rainfall events in the eastern monsoon region of China;

[0022] Figure 5 This is a temporal variation diagram of regional heavy rainfall events caused by tropical cyclones. The 95th threshold is used to identify regional heavy rainfall events.

[0023] Figure 6 This is a temporal variation diagram of regional heavy rainfall events caused by tropical cyclones. A 50mm threshold is used to identify regional heavy rainfall events.

[0024] Figure 7 This is a temporal variation diagram of regional heavy rainfall events in Northeast China caused by tropical cyclones;

[0025] Figure 8 This is a temporal variation diagram of regional heavy rainfall events in southeastern China caused by tropical cyclones;

[0026] Figure 9 is the QQ plot of the characteristics of regional heavy rainfall events caused by tropical cyclones, as observed and simulated by the generalized extreme value model;

[0027] Figure 10 is the QQ plot of the characteristics of regional heavy rainfall events caused by tropical cyclones, as observed and simulated by the generalized extreme value model;

[0028] Figure 11 This is a diagram showing the relationship between the characteristics of regional heavy rainfall events caused by tropical cyclones and tropical cyclones;

[0029] Figure 12 It is a working diagram of the hardware device of the present invention. DETAILED DESCRIPTION

[0030] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0031] Before formally explaining the present invention, the scheme of the present invention is first generally explained for easy understanding.

[0032] Please refer to Figure 1 , Figure 1 It is a schematic flow diagram of the method of the present invention.

[0033] The present invention provides a method for identifying heavy rainfall events in a tropical cyclone-induced region and analyzing non-steady-state characteristics, comprising the following steps:

[0034] S1. Data collection: Collect observation-based gridded precipitation data, tropical cyclone best track data, and meteorological reanalysis data;

[0035] As an exemplary embodiment, step S1 in the present invention is specifically as follows: the tropical cyclone optimal path data includes the latitude, longitude and maximum wind speed of the center position of the tropical cyclone once every 3 hours; the precipitation data is grid precipitation data with high temporal and spatial resolution, with a time resolution of 3 hours and a spatial resolution of 0.25°×0.25°; the meteorological reanalysis data includes wind fields, latitudinal and longitudinal water vapor transport.

[0036] S2. Identification of regional heavy rainfall events: Based on the definition of heavy rainfall and the method for determining regional heavy rainfall events, regional heavy rainfall events are identified in combination with the grid precipitation data collected in step S1;

[0037] As an exemplary embodiment, step S2 in the present invention is specifically as follows:

[0038] S21: The definition of heavy precipitation by the China Meteorological Administration, that is, the total amount of precipitation is not less than 50 mm, is used as the threshold for heavy precipitation defined in the present invention. Due to the large spatial variability of precipitation in the monsoon region, the 95th percentile (95th) of the non-zero precipitation value is also used as the threshold (critical value) for each grid. Grids whose daily precipitation values ​​exceed the threshold are extracted, and adjacent grids are merged into one event. In order to ensure that large-scale heavy precipitation events can cover a sufficiently large area, only events with an area greater than 10,000 square kilometers are considered;

[0039] S22: Method for determining regional heavy rainfall events - multi-temporal clustering algorithm; Heavy rainfall events can change continuously in time and space. It consists of multiple time states, and at least one heavy rainfall event will occur in each time state. However, for two adjacent time states, the distance moved by the same regional heavy rainfall event is very short, and there is overlap in space. Therefore, if the spatial topological structures of multiple heavy rainfall events in adjacent time states intersect (that is, they share an overlapping area and at least one grid overlaps), they are considered to belong to the same regional heavy rainfall event. These search processes will continue until there are no overlapping areas for regional heavy rainfall events. For each regional heavy rainfall event, its transfer, fission and fusion processes can be tracked in each time state, so that the identified regional heavy rainfall events have spatiotemporal continuity and a high degree of consistency with the facts.

[0040] S3. Correlation between tropical cyclones and regional heavy rainfall events: Based on the regional heavy rainfall events obtained in step S2, a multi-temporal clustering algorithm is used to obtain a mapping relationship between tropical cyclones and the regional heavy rainfall events they trigger.

[0041] As an exemplary embodiment, in step S3 of the present invention, a multi-temporal clustering algorithm is used to associate tropical cyclones with regional heavy rainfall events. The specific steps are as follows:

[0042] S31: Treat daily regional heavy precipitation events as rain bands;

[0043] S32: Calculate the distance between the tropical cyclone center and the weighted precipitation center of the regional heavy precipitation event (i.e., the grid point with the largest cumulative precipitation during the entire regional heavy precipitation event). If any distance is less than a certain threshold (e.g., 500 kilometers, which has been widely selected in previous studies), the regional heavy precipitation event is defined as a heavy precipitation event caused by the tropical cyclone. Multiple regional heavy precipitation events caused by the same tropical cyclone are merged into one event.

[0044] S33: Obtain a one-to-one mapping relationship between tropical cyclones and regional heavy rainfall events caused by tropical cyclones.

[0045] S4. Identification of characteristics of regional heavy rainfall caused by tropical cyclones: Based on the mapping relationship between tropical cyclones and regional heavy rainfall events caused by them obtained in step S3, the spatiotemporal evolution of the characteristics of regional heavy rainfall events caused by tropical cyclones is quantified from four aspects: duration, cumulative affected area, cumulative average precipitation depth, and cumulative total precipitation.

[0046] As an exemplary embodiment, step S4 of the present invention is specifically as follows:

[0047] S41: For each regional heavy rainfall event caused by a tropical cyclone, four indices are defined to evaluate its characteristics: duration, cumulative affected area, cumulative average rainfall depth, and cumulative total rainfall. The calculation formulas for the four indices are as follows:

[0048] (1) Duration (D; number of days): D = end date - start date + 1.

[0049] (2) Cumulative affected area (A; km 2 ), the sum of the daily affected areas during the entire duration:

[0050]

[0051] Among them, A t is the affected area on day t, and T is the total duration of the event.

[0052] (3) Cumulative mean precipitation depth (P; mm), the sum of the daily area-weighted mean precipitation depths of all affected grids during the entire duration:

[0053]

[0054] Among them, P t is the precipitation depth on day t, A t is the affected area on day t, A 总 It is the cumulative affected area during the entire duration.

[0055] (4) Total cumulative precipitation (V; km 3 ), the daily area-weighted mean precipitation depth for the entire duration multiplied by the sum of the daily affected areas:

[0056]

[0057] Among them, P t is the precipitation depth on day t, A t is the affected area on day t.

[0058] S42: The trend of the characteristics of regional heavy rainfall events caused by tropical cyclones is calculated by the least squares method. The slope obtained by fitting the time series of the study variables by the least squares method is the trend of change. The least squares fitting formula of the slope is as follows:

[0059]

[0060] Where b is the slope, n is the total number of years in the study period, x represents the year, and x represents the i is the i-th year in the study period, is the mean value of the research period; y represents the value of the research variable, y iis the value of the research variable in year i during the research period, is the mean value of the research variable during the research period.

[0061] S5. Analysis of the variability of regional heavy rainfall characteristics caused by tropical cyclones: combining the characteristics of regional heavy rainfall events caused by tropical cyclones obtained in step S4, quantifying the variability of event characteristics, and analyzing the mechanism affecting the characteristic variability;

[0062] As an exemplary embodiment, step S5 of the present invention is specifically as follows:

[0063] The maximum values ​​of the three variables of the cumulative affected area, cumulative average precipitation depth and cumulative total precipitation of tropical cyclone-induced regional heavy rainfall events in each year were extracted; the maximum values ​​of these three variables in each year were then sorted from large to small, and "Max10" refers to the data corresponding to the 10 years in the first 10 years of this sorting, and "Min10" refers to the data corresponding to the 10 years in the last 10 years in this sorting; based on the above two groups of 10-year data of these three variables (one group takes the maximum value and the other group takes the minimum value), the positive and negative anomalies of these three variables in the corresponding 10 years were calculated to analyze the mechanism of the characteristic variability that may affect the regional heavy rainfall caused by tropical cyclones.

[0064] S6. Non-steady-state frequency analysis of regional heavy rainfall characteristics caused by tropical cyclones: Based on the regional heavy rainfall events caused by tropical cyclones obtained in step S3, extreme value theory is used to estimate the recurrence period of the annual maximum cumulative impact area, cumulative area average precipitation depth, and cumulative total precipitation of regional heavy rainfall events in step S4, to test whether the probability of occurrence of extreme regional heavy rainfall events caused by tropical cyclones has changed;

[0065] As an exemplary embodiment, step S6 in the present invention is specifically as follows:

[0066] S61: Use Akaike Information Criterion to select the best model;

[0067] Akaike Information Criterion (AIC) is a statistic used for model selection. It aims to select the best statistical model by balancing the goodness of fit and model complexity. The calculation formula of Akaike Information Criterion (AIC) is as follows:

[0068] AIC=2k-2ln(L)

[0069] Where k is the number of parameters in the model. L is the maximum likelihood estimate of the model, which represents the probability of the data occurring given the given parameters. A lower AIC value indicates a better fit for the model, meaning the model is better.

[0070] S62: Select extreme value distribution model;

[0071] To assess the frequency of regional heavy rainfall events triggered by tropical cyclones, extreme value theory was used based on the Akaike Information Criterion to estimate the annual maximum recurrence period (i.e., occurring once every n years on average) for the cumulative affected area A, the cumulative mean precipitation depth P, and the cumulative total precipitation V. The formula for the cumulative distribution function of the generalized extreme value distribution is as follows:

[0072]

[0073] Where z is the time series of the annual maximum values ​​of the cumulative affected area A, the cumulative mean precipitation depth P, and the cumulative total precipitation V; μ, σ, and ξ are the location parameter, scale parameter, and shape parameter, respectively. If these three parameters remain constant, the generalized extreme value distribution is a stationary model. To establish a nonstationary model, the location parameter (μ) and scale parameter (σ) can be obtained as linear functions of the time-varying covariates:

[0074]

[0075] Where x is a time-varying covariate; μ0, σ0, α1, and β1 are regression parameters. μ0 represents the initial value of the location parameter, σ0 represents the initial value of the scale parameter, α1 represents the rate at which the location parameter changes with the independent variable, and β1 represents the rate at which the scale parameter changes with the independent variable. Linear regression analysis methods such as the least squares method are used to calculate estimated values ​​of μ0, σ0, α1, and β1, minimizing the error between the fitted line and the actual data. The shape parameter ξ is set to a constant due to the large uncertainty in its estimation. Therefore, three models were constructed: M0 (α1 = 0, β1 = 0); M1 (α1 ≠ 0, β1 = 0); and M2 (α1 ≠ 0, β1 ≠ 0). Model M0 is a steady-state model. Model M1 is a nonstationary model in which only the location parameter changes with time. Model M2, on the other hand, has both location and scale parameters that change with time.

[0076] To assess the changes in the frequency of regional heavy rainfall events caused by tropical cyclones, using time, tropical cyclone intensity, and calculated tropical cyclone movement speed from tropical cyclone data as covariates.

[0077] S63: Identify covariates;

[0078] The duration of a tropical cyclone's movement is the time difference between the first and last positions of a tropical cyclone within the study area, measured in hours. The speed of a tropical cyclone's movement is the average of all six-hour moving speeds of tropical cyclones within the study area. The six-hour speed is calculated by calculating the spherical distance between the centers of two consecutive tropical cyclones and dividing it by six hours. The angle deviation of a tropical cyclone's movement is defined as the average of the angle deviations of all consecutive six-hour path vectors of a tropical cyclone within the study area.

[0079] The calculation formula for the moving speed of a tropical cyclone is as follows:

[0080] L (n-1)n =R·arccos[cos(α n-1 -α n )cosβ n-1 cosβ n +sinβ n-1 sinβ n ]

[0081]

[0082] Since the time resolution of tropical cyclones is 6 hours, where n is the number of track positions of a tropical cyclone in the study area every 6 hours, L (n-1)n For tropical cyclones in p n-1 、p n The moving distance between two points, R is the radius of the earth, α n-1 , α n For p n-1 、p n Longitude of two points, β n-1 , β n For p n-1 、p n The latitude of two points, For tropical cyclones in p n-1 、p n The speed of movement between two points, is the moving speed of the tropical cyclone between points p1 and p2, is the moving speed of the tropical cyclone between points p2 and p3, is the moving speed of a tropical cyclone in the study area.

[0083] S64: Model validation;

[0084] Quantile-Quantile (QQ) plots can be used to compare the similarity of two distributions, thereby testing data normality or verifying model fit. In a QQ plot, the horizontal axis typically represents the quantiles of the theoretical distribution, and the vertical axis represents the quantiles of the actual data. If the simulated and observed quantiles follow a 1:1 straight line, the unsteady model provides a good fit to the observed regional heavy precipitation event index.

[0085] S7. Mechanism and impact of shortened return periods of extreme regional heavy rainfall events caused by tropical cyclones: Based on the changes in the characteristic return periods of regional heavy rainfall events caused by tropical cyclones obtained in step S6, analyze the factors affecting the shortened return period.

[0086] As an exemplary embodiment, in step S7 of the present invention: the Spearman correlation coefficient is used to analyze the correlation between the intensity and movement speed of tropical cyclones and regional heavy rainfall events caused by landfalling tropical cyclones; the significance of the obtained correlation is tested using a t-test; and the relationship between regional heavy rainfall events caused by tropical cyclones and tropical cyclones is analyzed based on the size of the correlation and its significance.

[0087] In order to better explain the present invention, a specific embodiment is listed below.

[0088] In this example, tropical cyclones that landed in China from 1961 to 2018 were used as an example. The study area was limited to the eastern monsoon region of China, and an objective recognition method combining a multi-temporal clustering algorithm and objective weather analysis technology was used to identify and study regional heavy rainfall events caused by tropical cyclones in this region.

[0089] The implementation flow chart of the method of the present invention is as follows Figure 1 The specific steps are as follows:

[0090] (1) Collection of basic data;

[0091] In this embodiment, the tropical cyclone best track data uses the tropical cyclone best track dataset of the International Climate Management Best Track Archive, which includes the cyclone center position and near-center maximum wind speed of tropical cyclones every 3 hours, and the dataset covers the period from 1961 to 2018. The precipitation data uses the gridded dataset (CN05.1) from 1961 to 2018 developed based on observation data from 2,416 meteorological stations across the country, with a temporal resolution of 3 hours and a horizontal spatial resolution of 0.25°×0.25°. The reanalysis data collects the ERA5 reanalysis data from the European Centre for Medium-Range Weather Forecasts (ECMWF), with a temporal resolution of monthly and a spatial resolution of 0.25°×0.25°. The data includes wind field, latitudinal water vapor flux, and meridional water vapor flux. The vertical wind shear is calculated based on the wind field between 200 hPa and 850 hPa, and the wind field at 500 hPa is used to guide the airflow.

[0092] (2) Identification of regional heavy rainfall events:

[0093] like Figure 2 As shown, the method of step S2 is used to identify and extract regional heavy rainfall events from the grid precipitation data obtained in step S1 according to the definition of heavy rainfall and the multi-temporal clustering algorithm.

[0094] The main process of the multi-temporal clustering algorithm for identifying regional heavy rainfall events includes: setting the 95th percentile of daily precipitation of not less than 50 mm and non-zero precipitation values ​​as thresholds to define heavy rainfall respectively, identifying heavy rainfall events, and extracting grid points that exceed the set thresholds in each time state. Only adjacent grid points exceeding the threshold are considered to be merged into an area larger than 10,000 square kilometers to ensure that the coverage of heavy rainfall events is large enough; based on the spatial topological intersection of multiple heavy rainfall events in adjacent time states, it is identified whether they belong to the same regional heavy rainfall event. If two events overlap in space (that is, at least one grid point overlaps), they are considered to be the same event. This process is repeated until there is no longer any event overlap; for each regional heavy rainfall event, its transmission, fission, and fusion are tracked in each time state.

[0095] (3) Association between tropical cyclones and regional heavy rainfall events:

[0096] like Figure 2As shown, the method of step S3 is adopted to associate the tropical cyclone with the regional heavy rainfall event obtained in step S2. First, the regional heavy rainfall event of each day is regarded as a rain belt, and the distance between the center of the tropical cyclone and the weighted precipitation center of the regional heavy rainfall event (i.e., the grid point with the largest cumulative precipitation during the entire regional heavy rainfall event) is calculated. If any distance is less than 500 kilometers, the regional heavy rainfall event is identified as an event caused by the tropical cyclone center. Multiple regional heavy rainfall events caused by the same tropical cyclone center are merged into one event. Finally, a mapping relationship between the tropical cyclone and the regional heavy rainfall event guided by the tropical cyclone is obtained.

[0097] For each regional heavy rainfall event caused by a tropical cyclone, four indices are used to evaluate its characteristics. Figure 2 , D, A, P and V refer to the duration, cumulative affected area, cumulative average precipitation depth and cumulative total precipitation of regional heavy rainfall events caused by tropical cyclones, respectively.

[0098] Based on this identification framework, if 50 mm / 95th is used as the critical value, this example extracts 472 / 571 regional heavy rainfall events caused by tropical cyclones in the eastern China monsoon region between 1961 and 2018.

[0099] (4) Identification of the characteristics of regional heavy rainfall caused by tropical cyclones:

[0100] like Figures 3 to 8 As shown in the figure, the method of step S4 is used to calculate the four characteristic indices of each regional heavy rainfall event: duration (D; number of days), cumulative affected area (A; km 2 ), cumulative average precipitation depth (P; mm), cumulative total precipitation (V; km 3 The time series of the characteristics (four indices) of regional heavy rainfall events caused by tropical cyclones were statistically analyzed, and the changing trend of the characteristics of regional heavy rainfall events caused by tropical cyclones was calculated by the least squares method. The slope obtained by fitting the time series of the research variables by the least squares method is the changing trend.

[0101] like Figure 3 As shown in Figure 2, the four characteristics of regional heavy rainfall events caused by tropical cyclones all show a skewed distribution. Specifically, about 92.5% / 77.9% of the total events lasted no more than 4 days. Although 88.3% / 63.0% of the events had a cumulative impact area of ​​no more than 20×10 4 km 2 However, the maximum cumulative impact area of ​​regional heavy rainfall events caused by tropical cyclones reached 192.7×10 4 km 2The cumulative average precipitation depth of events did not exceed 200 mm in 53.2% / 55.3%, while the maximum cumulative average precipitation depth reached 1139.5 mm. Similarly, the cumulative total precipitation of regional heavy rainfall events caused by tropical cyclones did not exceed 20 km in 83.7% / 67.1%. 3 , but its maximum value is 246.7km 3 These results indicate that the spatial extent and total precipitation of regional heavy rainfall events induced by tropical cyclones exhibit significant heterogeneity among these events.

[0102] Figure 4 The seasonal characteristics of regional heavy rainfall events caused by tropical cyclones are shown. In this embodiment, tropical cyclones within a radius of 500 kilometers from the coastline were counted, and it was found that 79.3% of tropical cyclones occurred from June to September. Correspondingly, during the period from June to September, the characteristic values ​​of regional heavy rainfall events caused by tropical cyclones were 86.7% / 88.5% events, 87.7% / 91.1% cumulative affected areas, 86.0% / 86.6% cumulative average precipitation depths, and 87.2% / 89.5% cumulative total precipitation. These characteristics reached their peaks from June to September, with the highest values ​​in July. If attention is turned to the characteristics of regional heavy rainfall events caused by a single tropical cyclone, the inverted "V" shape of the seasonal distribution becomes flatter. The cumulative affected area, cumulative average precipitation depth, and cumulative total precipitation of regional heavy rainfall events of a single tropical cyclone increased sharply to about 10×10 4 km 2 , 80mm and 6km 3 The seasonality of heavy rainfall events triggered by a single tropical cyclone shows weak seasonality, which poses a greater challenge to the prediction and prevention of regional heavy rainfall disasters caused by tropical cyclones.

[0103] Spatially, the incidence of regional heavy precipitation events caused by tropical cyclones gradually decreases from the coast toward the inland. Southeast China experiences the most frequent regional heavy precipitation events (an average of 86.8 / 96.1 events). When using the 95th threshold, the areas with high and long duration tropical cyclone-induced regional heavy precipitation events are significantly greater than the 50 mm threshold. However, within the EMC region, the contribution of tropical cyclone-induced regional heavy precipitation events to the total regional heavy precipitation events is essentially the same under both thresholds, with the exception of Northeast China. In the southeastern coastal region, the contribution rates under the two thresholds are 48.4% and 42.4%, respectively, which are almost equal to the proportion of tropical cyclone-induced localized heavy precipitation events. In Northeast China, when using the 95th threshold, tropical cyclone-induced regional heavy precipitation events account for less than 20% of the total events, while using the 50 mm threshold, this proportion exceeds 40%. This suggests that tropical cyclones play a significant role in the widespread heavy rainfall exceeding 50 mm and the subsequent widespread flooding in Northeast China.

[0104] The characteristics of these tropical cyclone-induced regional heavy rainfall events are particularly pronounced in southeastern China, with a significant increase in frequency and duration. The trends in Northeast China differ significantly between the two thresholds for tropical cyclone-induced regional heavy rainfall events. When the 50mm threshold is used, the frequency and duration of tropical cyclone-induced regional heavy rainfall events in Northeast China increase, while the opposite is true when the 95mm threshold is used.

[0105] Figure 5 The temporal variation characteristics of regional heavy rainfall events caused by tropical cyclones, with the 95th threshold, are shown. The annual frequency and duration of various characteristics of regional heavy rainfall events caused by tropical cyclones show a slight downward trend, with decreases of -0.59% / decade and -1.16% / decade, respectively. The cumulative impact area also shows a downward trend (-2.72% / decade). However, the cumulative average precipitation depth and cumulative total precipitation both show an upward trend, with increases of 1.22% / decade and 1.12% / decade, respectively. The trend signals of all characteristics of regional heavy rainfall events caused by tropical cyclones are not strong and stable, as none of them pass the statistical significance test. The inter-decadal variation rates of these characteristics with a period of approximately 20 years are more significant. Overall, for all characteristics of regional heavy rainfall events triggered by tropical cyclones, this interdecadal variation shows a consistent cycle of decrease and increase, namely, decrease from 1961 to 1980, increase from 1980 to 1995, decrease from 1995 to 2005, and increase again from 2005 to 2019.

[0106] Figure 6The data show the temporal variation of regional heavy rainfall events caused by tropical cyclones, with a threshold of 50 mm. All characteristics of these events have increased. The frequency, duration, cumulative affected area, cumulative average rainfall depth, and total rainfall have increased at rates of 1.56% per decade, 3.15% per decade, 5.15% per decade, 3.41% per decade, and 5.45% per decade, respectively. Regional heavy rainfall events exceeding 50 mm caused by tropical cyclones primarily occur in southeastern China, resulting in more frequent, more extensive, and more intense large-scale floods in this region.

[0107] Figures 7-8 The authors focused on the changes in the characteristics of regional heavy rainfall events caused by tropical cyclones in Northeast China and the southeastern coastal areas. From 1961 to 2018, the regional heavy rainfall event index caused by tropical cyclones in Northeast China showed a downward trend ( Figure 7 ), which is consistent with Figure 5 The significant downward trend in frequency and duration shown in c and d is consistent with that in the previous section. In the southeastern coastal area of ​​China, the regional heavy rainfall event indices caused by tropical cyclones mostly show positive trends, although these positive trends are not as obvious as the negative trends in northeastern China ( Figure 8 The differences in the trends of regional heavy rainfall events caused by tropical cyclones in Northeast China and the southeastern coastal areas indicate that the risks associated with regional heavy rainfall events caused by tropical cyclones have different regional patterns.

[0108] (5) Analysis of the mechanism of the characteristic variability of regional heavy rainfall caused by tropical cyclones:

[0109] As shown in Tables 1 and 2, the method of step S5 is used to analyze the mechanism of the characteristic variability of regional heavy rainfall caused by tropical cyclones. In this embodiment, two groups of 10-year data of regional heavy rainfall events caused by 472 / 571 tropical cyclones in the eastern monsoon region of China from 1961 to 2018 are selected to explore the mechanism of the variability. One group takes the maximum value and the other group takes the minimum value (see Table 1). The recognition threshold of regional heavy rainfall events is 50mm / 95th; based on the above two groups of 10-year data, this embodiment constructs a table of 10-year positive and negative anomalies (see Table 2), and compares and analyzes the composite maps of positive and negative anomalies.

[0110] Comparison and analysis reveal interannual variations in 850hPa horizontal winds, vertical wind shear, guiding air currents, and integrated water vapor transport from June to September. These large-scale environmental anomalies have been shown to strongly influence the frequency, duration, and path of hot cyclones. During the decade with positive anomalies, prevailing easterly winds over the western Pacific warm pool and the Chinese landmass pushed the Pacific warm pool westward, thereby shifting the tropical cyclone genesis region westward. Consequently, these prevailing easterly winds provide favorable dynamic conditions for tropical cyclones to develop toward eastern China. Simultaneously, westward guiding air currents over the central and eastern Pacific contribute to the landward movement of typhoons. An anomaly-induced guiding air current developed over southeastern China, favoring the passage of more tropical cyclones through the country. Southeasterly winds blowing east of the monsoon trough form guiding air currents that guide tropical cyclones toward landfall along China's coastal areas. In years with positive anomalies, vertical wind shear also weakens, favoring the formation and maintenance of tropical cyclone warm cores and, consequently, the maintenance of tropical cyclone intensity after landfall. In positive anomaly years, integrated water vapor transport is characterized by a water vapor flux path from the western Pacific Ocean (WNP), the South China Sea, and the Bay of Bengal to southeastern China. In particular, an anomaly anticyclone is observed near the eastern waters of China, enhancing water vapor transport from the western Pacific Ocean to land. This intensive water vapor flux provides an abundant water vapor supply. On the one hand, warm and moist airflow can provide energy to maintain the intensity of tropical cyclones, and on the other hand, it provides favorable conditions for heavy rainfall caused by tropical cyclones. In negative anomaly years, this atmospheric circulation pattern is not observed or displays opposite characteristics.

[0111] Table 1: Maximum and minimum values ​​of duration (D), cumulative affected area (A), cumulative average precipitation depth (P), and cumulative total precipitation (V) of regional heavy rainfall events caused by tropical cyclones in the past 10 years

[0112]

[0113] Note: The first / second values ​​of " / " are used for regional heavy rainfall events caused by tropical cyclones, with a threshold of 50 mm / 95th.

[0114] Table 2: Anomalies of the two data sets in Table 2. The last row is the standard deviation of the entire time series of these indices for regional heavy rainfall events caused by tropical cyclones.

[0115]

[0116] Note: The first / second values ​​of " / " are used for regional heavy rainfall events caused by tropical cyclones, with a threshold of 50 mm / 95th.

[0117] (6) Non-stationary frequency analysis of the characteristics of regional heavy rainfall caused by tropical cyclones;

[0118] like Figures 9-10 As shown in Table 3, the generalized extreme value theory in step S6 is used to estimate the extreme values ​​of the four indices that characterize regional heavy rainfall events caused by tropical cyclones within a fixed return period, clarifying the temporal variation of the probability and intensity of extreme regional heavy rainfall events caused by tropical cyclones. The specific steps are as follows:

[0119] 1) Data preparation:

[0120] Extract the maximum values ​​of the three variables (58 data points in total) for each year of the cumulative affected area, cumulative average precipitation depth, and cumulative total precipitation of regional heavy rainfall events caused by tropical cyclones from 1961 to 2018 obtained in step S5;

[0121] 2) Select the extreme value distribution model:

[0122] In this example, a generalized extreme value distribution is used to describe annual extreme value data. The model parameters are set to change over time, and a non-stationary model of the time series of three characteristic indices of regional heavy rainfall events (i.e., cumulative affected area A, cumulative average precipitation depth P, and cumulative total precipitation V) is established. The location parameter and scale parameter change over time, μ = a0 + a1x, σ = β0 + β1x, and the shape parameter remains unchanged. In this embodiment, the time, tropical cyclone intensity, and tropical cyclone movement speed obtained by calculation in the tropical cyclone data of the International Climate Management Optimal Path Archive are used as covariates;

[0123] 3) Model Validation:

[0124] This example uses a quantile-quantile (QQ) graph to evaluate the goodness of fit. Figure 9 and Figure 10 As shown in the figure, the X-axis is the quantile of the regional heavy rainfall event index simulated by the optimal non-steady-state model, and the Y-axis is the quantile of the observed regional heavy rainfall event index. Figure 9 and Figure 10 It can be seen that the quantiles of the observed and simulated values ​​are almost on the 1:1 straight line, indicating that the simulation results of the optimal model are in good agreement with the observations, and the generalized extreme value distribution can well fit the annual maximum time series of regional heavy rainfall events caused by tropical cyclones;

[0125] 4) Find the extreme value within the specified return period:

[0126] Dividing the 58 years into two 29-year time periods (1961-1989 and 1990-2018), the generalized extreme value distribution (GEV) was used to estimate the 10-, 20-, and 50-year extremes (i.e., return periods of 10, 20, and 50 years) of the cumulative impact area, cumulative average precipitation depth, and cumulative total precipitation of heavy rainfall events caused by tropical cyclones in the two sub-periods. The return period level was inferred from the GEVD function:

[0127]

[0128] Compared to the period 1961–1989, regional heavy precipitation events triggered by tropical cyclones over the past 29 years have shown significant increases in all three of the above characteristics (see Table 3). For example, the cumulative affected area, cumulative mean precipitation depth, and cumulative total precipitation over the 20-year period increased by 58.9% / 15.5%, 16.4% / 30.1%, and 59.0% / 46.0%, respectively. Furthermore, these increases were even more pronounced for more extreme regional heavy precipitation events triggered by tropical cyclones. Specifically, for the 50-year period, these percentages were 83.7% / 15.2%, 16.1% / 39.4%, and 76.3% / 58.3%, respectively. The increase in extreme values ​​of the regional heavy precipitation event index within a fixed decadal period also indicates a shortening of the return period of extreme, regionally specific heavy precipitation events triggered by tropical cyclones.

[0129] Table 3: Changes in the cumulative impact area (A), cumulative mean precipitation depth (P), and cumulative total precipitation (V) of TC-induced RHPEs within 10-, 20-, and 50-year return periods

[0130]

[0131] Note: The first / second values ​​of " / " apply to regional heavy rainfall events caused by tropical cyclones, with a threshold of 50 mm / 95th.

[0132] (7) Mechanisms and impacts of shortened return periods of extreme regional heavy rainfall events caused by tropical cyclones:

[0133] In this example, tropical cyclones are classified according to their intensity into tropical depression, tropical storm, severe tropical storm, typhoon, severe typhoon and super typhoon. The median of the tropical cyclone intensity and the regional heavy rainfall event characteristics caused by tropical cyclones in each category are extracted and paired ( Figure 11 ac). Tropical cyclones are divided into ten quantiles according to their moving speed. In every two adjacent quantiles, the median of the tropical cyclone moving speed and the regional heavy rainfall event characteristics caused by the tropical cyclone are extracted and paired ( Figure 11 df).

[0134] As the intensity of tropical cyclones increases, the three characteristic indices of regional heavy rainfall events caused by tropical cyclones will become larger, showing a positive correlation. The situation is similar for the speed of tropical cyclones, but when the speed of tropical cyclones is faster, the cumulative average precipitation depth becomes smaller ( Figure 11 e) It is worth noting that the characteristics of regional heavy rainfall events caused by tropical cyclones are not linearly related to the intensity and movement speed of the tropical cyclones.

[0135] This example further uses a non-stationary model to explore the impact of tropical cyclone intensity and speed on the probability of regional heavy rainfall events. The 10 years with the largest and 10 years with the smallest tropical cyclone intensity and speed are introduced into the model as covariates. When the intensity or speed is fixed, the impact of the other variable on the probability of regional heavy rainfall events is analyzed. As shown in Table 3, the probability of occurrence of high-intensity tropical cyclone and speed events is significantly different from that of low-intensity tropical cyclone and speed events. Figure 11 The results are generally consistent. For example, when tropical cyclones move faster, the 50-year cumulative impact area and cumulative total precipitation increase by 2.2% / 3.4% and 1.5% / 13.4%, respectively, while also reducing the 50-year cumulative average precipitation depth by -5.5% / -9.6%. Intense tropical cyclones also increase the 50-year cumulative impact area and cumulative total precipitation by 1.6% / 6.9% and 2.5% / 1.7%, respectively. Numerical simulations from climate models predict that under anthropogenic warming, tropical cyclones will slow down their movement and significantly increase their intensity. This suggests that the eastern monsoon region of China is more likely to be hit by rare, destructive, regional heavy precipitation events triggered by tropical cyclones in the future.

[0136] See Figure 12 , Figure 12 4 is a schematic diagram of the working of the hardware device of an embodiment of the present invention, wherein the hardware device specifically comprises: a device 401 for identifying and analyzing non-steady-state characteristics of heavy rainfall events in a tropical cyclone-induced region, a processor 402 and a storage medium 403.

[0137] A device 401 for identifying and analyzing non-steady-state characteristics of heavy rainfall events in a tropical cyclone-induced region: The device 401 for identifying and analyzing non-steady-state characteristics of heavy rainfall events in a tropical cyclone-induced region implements a method for identifying and analyzing non-steady-state characteristics of heavy rainfall events in a tropical cyclone-induced region.

[0138] Processor 402: The processor 402 loads and executes the instructions and data in the storage medium 403 to implement the method for identifying heavy rainfall events in a tropical cyclone-induced area and analyzing non-steady-state characteristics.

[0139] Storage medium 403: The storage medium 403 stores instructions and data; the storage medium 403 is used to implement the method for identifying heavy rainfall events in a tropical cyclone-induced area and analyzing non-steady-state characteristics.

[0140] The beneficial effects of the present invention are:

[0141] (1) The present invention discloses a method for identifying regional heavy rainfall events caused by tropical cyclones, determines the mapping relationship between tropical cyclones and regional heavy rainfall events caused by tropical cyclones, and quantifies the spatiotemporal evolution characteristics of this type of heavy rainfall events in the historical period from four aspects: duration, cumulative affected area, cumulative average precipitation depth and cumulative total precipitation. The method of the present invention can fully identify the spatiotemporal evolution and movement characteristics of regional extreme precipitation caused by tropical cyclones, which will further improve the accuracy of heavy rainfall forecasts caused by tropical cyclones.

[0142] (2) The non-steady-state characteristics of regional heavy rainfall events caused by tropical cyclones were analyzed. The non-steady-state frequency of extreme regional heavy rainfall events caused by tropical cyclones was quantified from four aspects: duration, cumulative affected area, cumulative average precipitation depth and cumulative total precipitation. The extreme value theory was used to predict the future recurrence period of extreme regional heavy rainfall events caused by tropical cyclones, and the mechanism and impact of the shortening of the recurrence period were explored, providing a scientific basis for resisting disasters caused by tropical cyclones and making decisions.

[0143] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for identifying heavy rainfall events in a tropical cyclone-induced region and analyzing their non-steady-state characteristics, characterized by: The method comprises the following steps: S1. Data collection: Collect observation-based gridded precipitation data, tropical cyclone best track data, and meteorological reanalysis data; S2. Identification of regional heavy rainfall events: Based on the definition of heavy rainfall and the method for determining regional heavy rainfall events, regional heavy rainfall events are identified in combination with the grid precipitation data collected in step S1; S3. Correlation between tropical cyclones and regional heavy rainfall events: Based on the regional heavy rainfall events obtained in step S2, a multi-temporal clustering algorithm is used to obtain a mapping relationship between tropical cyclones and the regional heavy rainfall events they trigger. S4. Identification of characteristics of regional heavy rainfall caused by tropical cyclones: Based on the mapping relationship between tropical cyclones and regional heavy rainfall events caused by them obtained in step S3, the spatiotemporal evolution of the characteristics of regional heavy rainfall events caused by tropical cyclones is quantified from four aspects: duration, cumulative affected area, cumulative average precipitation depth, and cumulative total precipitation. S5. Analysis of the variability of regional heavy rainfall characteristics caused by tropical cyclones: combining the characteristics of regional heavy rainfall events caused by tropical cyclones obtained in step S4, quantifying the variability of event characteristics, and analyzing the mechanism affecting the characteristic variability; S6. Non-steady-state frequency analysis of regional heavy rainfall characteristics caused by tropical cyclones: Based on the regional heavy rainfall events caused by tropical cyclones obtained in step S3, extreme value theory is used to estimate the recurrence period of the annual maximum cumulative impact area, cumulative area average precipitation depth, and cumulative total precipitation of regional heavy rainfall events in step S4, to test whether the probability of occurrence of extreme regional heavy rainfall events caused by tropical cyclones has changed; S7. Mechanism and impact of shortened return periods of extreme regional heavy rainfall events caused by tropical cyclones: Based on the changes in the characteristic return periods of regional heavy rainfall events caused by tropical cyclones obtained in step S6, analyze the factors affecting the shortened return period.

2. The method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region according to claim 1, wherein: Step S1 is specifically as follows: The tropical cyclone optimal path data includes the latitude, longitude and maximum wind speed of the center of the tropical cyclone every 3 hours; the precipitation data is grid precipitation data with high temporal and spatial resolution, with a time resolution of 3 hours and a spatial resolution of 0.25°×0.25°; the meteorological reanalysis data includes wind fields, latitudinal and longitudinal water vapor transport.

3. The method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region according to claim 2, wherein: Step S2 is specifically as follows: S21: Heavy rainfall refers to a total amount of rainfall of not less than 50 mm. Grids with daily precipitation values ​​exceeding a preset threshold are extracted, and adjacent grids are merged into one event. S22: Identify regional heavy precipitation events using a multi-temporal clustering algorithm.

4. The method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region according to claim 3, wherein: Step S3 is as follows: A multi-temporal clustering algorithm is used to associate tropical cyclones with regional heavy rainfall events. The specific steps are as follows: S31: Treat daily regional heavy precipitation events as rain bands; S32: Calculate the distance between the tropical cyclone center and the weighted precipitation center of the regional heavy precipitation event; if any distance is less than a certain threshold, the regional heavy precipitation event is defined as a heavy precipitation event caused by the tropical cyclone; multiple regional heavy precipitation events caused by the same tropical cyclone are merged into one event; S33: Obtain a one-to-one mapping relationship between tropical cyclones and regional heavy rainfall events caused by tropical cyclones.

5. The method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region according to claim 4, characterized in that: Step S4 is specifically as follows: S41: For each regional heavy rainfall event caused by a tropical cyclone, four indices are defined to assess its characteristics: duration, cumulative affected area, cumulative average rainfall depth, and cumulative total rainfall. The calculation formulas for the four indices are as follows: (1) Duration (D; number of days): D = end date - start date + 1; (2) Cumulative affected area, the sum of the affected areas every day during the entire duration: Among them, A t is the affected area on day t, and T is the total duration of the event; (3) Cumulative average precipitation depth, the sum of the daily area-weighted average precipitation depths of all affected grids during the entire duration: Among them, P t is the precipitation depth on day t, A t is the affected area on day t, A 总 is the cumulative affected area over the entire duration; (4) Cumulative total precipitation, which is the sum of the daily area-weighted average precipitation depth during the entire duration multiplied by the daily affected area: Among them, P t is the precipitation depth on day t, A t is the affected area on day t; S42: The trend of the characteristics of regional heavy rainfall events caused by tropical cyclones is calculated by the least squares method. The slope obtained by fitting the time series of the study variables by the least squares method is the trend of change. The least squares fitting formula of the slope is as follows: Where b is the slope, n is the total number of years in the study period, x represents the year, and x represents the i is the i-th year in the study period, is the mean value of the research period; y represents the value of the research variable, y i is the value of the research variable in year i during the research period, is the mean value of the research variable during the research period.

6. The method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region according to claim 5, characterized in that: Step S5 is specifically as follows: The maximum values ​​of the three variables of the cumulative affected area, cumulative average precipitation depth, and cumulative total precipitation in each year of tropical cyclone-induced regional heavy rainfall events were extracted. The maximum values ​​of these three variables in each year were then sorted from large to small, with Max10 representing the data corresponding to the 10 years in the first 10 years of this sorting, and Min10 representing the data corresponding to the 10 years in the last 10 years of this sorting. Based on the above two groups of 10-year data of these three variables, one group took the maximum value and the other group took the minimum value, and the positive and negative anomalies of these three variables in the corresponding 10 years were calculated to analyze the mechanism of the characteristic variability that may affect regional heavy rainfall caused by tropical cyclones.

7. The method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region according to claim 6, characterized in that: Step S6 is specifically as follows: S61: Use Akaike Information Criterion to select the best model; S62: Select the extreme value distribution model; use the extreme value theory based on the Akaike Information Criterion to estimate the return period of the annual maximum values ​​of the cumulative affected area A, the cumulative average precipitation depth P, and the cumulative total precipitation V. The formula for the cumulative distribution function of the generalized extreme value distribution is as follows: Where z is the annual maximum time series of the cumulative affected area A, the cumulative mean precipitation depth P, and the cumulative total precipitation V; μ, σ, and ξ are the location parameter, scale parameter, and shape parameter, respectively. If these three parameters remain unchanged, the generalized extreme value distribution is a steady-state model. To establish a non-stationary model, the location parameter (μ) and scale parameter (σ) are obtained as linear functions of the time-varying covariates: Among them, x is a covariate that changes with time; μ0, σ0, α1, and β1 are regression parameters, μ0 represents the initial value of the location parameter, σ0 represents the initial value of the scale parameter, α1 represents the rate at which the location parameter changes with the independent variable, and β1 represents the rate at which the scale parameter changes with the independent variable. Linear regression analysis methods such as the least squares method are used to calculate the estimated values ​​of μ0, σ0, α1, and β1, so that the error between the fitted straight line and the actual data is minimized; the shape parameter ξ is set to a constant due to the large uncertainty in its estimation; therefore, three models are constructed: model M0, with parameters α1=0, β1=0; model M1 with parameters α1≠0, β1=0; and model M2 with parameters α1≠0, β1≠0; model M0 is a steady-state model; model M1 is a non-stationary model, in which only the location parameter changes with time; and model M2 with both location and scale parameters changing with time; Assess the changes in the frequency of regional heavy rainfall events caused by tropical cyclones, using time, tropical cyclone intensity, and calculated tropical cyclone speed from tropical cyclone data as covariates; S63: Identify covariates; The duration of a tropical cyclone's movement is the time difference between the first and last positions of a tropical cyclone in the study area, measured in hours. The moving speed of a tropical cyclone is the average of all 6-hour moving speeds of a tropical cyclone in the study area. The 6-hour moving speed is obtained by calculating the spherical distance between the centers of two consecutive tropical cyclones and dividing it by 6 hours. The angular deviation of a tropical cyclone is defined as the average of the angular deviations of all consecutive 6-hour path vectors of a tropical cyclone in the study area. The calculation formula for the moving speed of a tropical cyclone is as follows: L (n-1)n =R arccos[cos(α n-1 -a n )cosβ n-1 cosβ n +sinβ n-1 sinβ n ] Since the time resolution of tropical cyclones is 6 hours, where n is the number of track positions of a tropical cyclone in the study area every 6 hours, L (n-1)n For tropical cyclones in p n-1 、p n The moving distance between two points, R is the radius of the earth, α n-1 , α n For p n-1 、p n The longitude of two points, β n-1 , β n For p n-1 、p n The latitude of two points, For tropical cyclones in p n-1 、p n The speed of movement between two points, is the moving speed of the tropical cyclone between points p1 and p2, is the moving speed of the tropical cyclone between points p2 and p3, is the moving speed of a tropical cyclone in the study area; S64: Model validation; The quantile-quantile QQ plot is used to test the normality of the data or verify the fitting effect of the model. In the QQ plot, the horizontal axis represents the quantile of the theoretical distribution, and the vertical axis represents the quantile of the actual data. If the simulated quantile and the observed quantile are distributed along a 1:1 straight line, it means that the non-steady-state model can fit the observed regional heavy rainfall event index.

8. The method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region according to claim 7, characterized in that: In step S7: the Spearman correlation coefficient is used to analyze the correlation between the intensity and movement speed of tropical cyclones and the regional heavy rainfall events caused by landfalling tropical cyclones; The significance of the correlation obtained by t-test was used; the relationship between regional heavy rainfall events caused by tropical cyclones and tropical cyclones was analyzed based on the size of the correlation and its significance.

9. A storage medium, characterized in that: The storage medium stores instructions and data for implementing a method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region as described in any one of claims 1 to 8.

10. A device for identifying heavy rainfall events in a tropical cyclone-induced region and analyzing non-steady-state characteristics, characterized by: include: A processor and a storage medium; the processor loads and executes instructions and data in the storage medium to implement a method for identifying heavy rainfall events and analyzing non-steady-state characteristics in a tropical cyclone-induced region as described in any one of claims 1 to 8.

Images