Search

CN-122021238-A - Multi-disaster nonlinear and non-stationary joint probability distribution modeling method considering climate change influence

CN122021238ACN 122021238 ACN122021238 ACN 122021238ACN-122021238-A

Abstract

The invention belongs to the technical field of engineering disaster prevention and reduction, and relates to a multi-disaster nonlinear and non-stationary joint probability distribution modeling method considering influence of climate change. The method comprises the first step of carrying out trend analysis on the statistical annual maximum value data and initially identifying time-varying characteristics; the method comprises the steps of carrying out periodic analysis on statistical annual maximum data, identifying periodic components hidden in the data and obtaining dominant frequency, carrying out nonlinear edge distribution modeling on the annual maximum data by adopting a nonlinear extremum distribution model and calculating a time-varying reproduction period value, and carrying out nonlinear joint distribution modeling on the annual maximum data by adopting a nonlinear copula model and calculating a time-varying joint reproduction period contour line. The method provided by the invention provides a stronger analysis framework which is more in line with physical reality for understanding and quantifying the composite disaster risk in the climate change era, and has great theoretical value and wide application prospect.

Inventors

  • SHENG CHAO
  • ZHANG YUANBO
  • Dai kaoshan

Assignees

  • 四川大学

Dates

Publication Date
20260512
Application Date
20251215

Claims (7)

  1. 1. A multi-disaster nonlinear non-stationary joint probability distribution modeling method considering climate change influence is characterized by comprising the following steps: The first step, carrying out trend analysis on the statistical annual maximum value data, and primarily identifying time-varying characteristics; secondly, periodically analyzing the statistical annual maximum value data, identifying the periodic components hidden in the data, and obtaining dominant frequency; Thirdly, modeling nonlinear and non-stationary edge distribution of annual maximum data by adopting a non-stationary extremum distribution model, and calculating a time-varying reproduction period value; And fourthly, modeling nonlinear and non-stationary joint distribution of the annual maximum data by adopting a non-stationary copula model, and calculating a time-varying joint reproduction period contour line.
  2. 2. The modeling method of multi-disaster nonlinear non-stationary joint probability distribution considering climate change influence as claimed in claim 1, wherein in the first step, a sliding time window method is adopted to conduct trend analysis, the time length and sliding step size of the window are firstly determined, then the mean value, standard deviation and correlation coefficient of data in the window are calculated from the front end of the annual maximum data sequence, then the window is slid backwards according to the sliding step length, the process is repeated until the end of the data sequence is reached, and the significance level is given The mean value, standard deviation and correlation coefficient sequences obtained by sliding time window calculation are subjected to trend test by adopting a linear regression or M-K test method, and the test is calculated Value of if The sequence is shown to have a significant trend, i.e., an immediate variable characteristic.
  3. 3. The modeling method of multi-disaster nonlinear non-stationary joint probability distribution considering climate change influence according to claim 1 is characterized in that in the second step, a hilbert-yellow transformation is adopted to conduct periodic analysis, first empirical mode decomposition is conducted on a data sequence with a maximum annual value to obtain intrinsic modes of the sequence, then hilbert transformation is conducted on each intrinsic mode, a hilbert spectrum is obtained through grouping, finally a marginal spectrum is obtained through integration of the hilbert spectrum in time, and frequencies corresponding to significant amplitude values are selected from the marginal spectrum to serve as dominant frequencies.
  4. 4. The modeling method of multi-disaster nonlinear non-stationary joint probability distribution considering climate change influence according to claim 1, wherein in the third step, nonlinear non-stationary edge distribution modeling is performed on the annual maximum data sequence by adopting a non-stationary generalized extremum distribution model based on a generalized additive model: ; ; In the formula, In order to be able to take time, And Nonlinear predictors of the position parameter and the scale parameter respectively, Representing a smoothing function of the generalized additive model, To smooth the dimension of the term, i.e. the effective degree of freedom, Is the dominant frequency.
  5. 5. The method for modeling a multi-disaster nonlinear non-stationary joint probability distribution considering climate change effects as claimed in claim 4, wherein for time-varying position parameters Adopting an identical form of connection function, and time-varying scale parameters A logarithmic form of the join function is employed: ; 。
  6. 6. the modeling method of multi-disaster nonlinear non-stationary joint probability distribution considering influence of climate change according to claim 4, wherein for the non-stationary generalized extremum distribution model obtained by modeling, the time-varying reproduction period is calculated by adopting the following formula: ; In the formula, Is the first A non-stationary generalized extremum distribution of the year, Is the first The value of the reproduction period of the year, Is the first A period of reproduction of the year; The average time interval of the effective data in the annual maximum data sequence is represented, namely the total years divided by the number of the effective data; The parameters of GEV distribution are respectively position parameters, scale parameters and shape parameters.
  7. 7. The modeling method of multi-disaster nonlinear and non-stationary joint probability distribution considering climate change influence according to claim 1, wherein in the fourth step, nonlinear and non-stationary joint distribution modeling is performed on the annual maximum data sequences related to each other by adopting a non-stationary copula model based on a generalized additive model, a binary copula model is adopted for two-dimensional data, and a nested copula model is adopted for high-dimensional data.

Description

Multi-disaster nonlinear and non-stationary joint probability distribution modeling method considering climate change influence Technical Field The invention belongs to the technical field of engineering disaster prevention and reduction, and particularly relates to a multi-disaster nonlinear and non-stationary joint probability distribution modeling method considering influence of climate change. Background The global air temperature continues to rise, resulting in significant increases in the frequency, intensity and complexity of extreme climates and weather events. This is particularly manifested by extra large floods, extreme drought, super strong typhoons, hot waves, storm surge and secondary disasters caused thereby. And these natural disasters are often not isolated events, and their interactions and chain reactions constitute a complex "multi-disaster" risk system. Therefore, the method for scientifically and accurately evaluating the multi-disaster combined risk under global climate change is one of urgent demands and key problems for improving social toughness and disaster prevention and reduction capability. However, conventional multi-disaster assessment methods based on the stationarity assumption expose a great limitation in the face of the strong driving force of climate change. The traditional method considers that the statistical rules (such as mean value, variance and extremum distribution) of disaster events are kept unchanged in the time dimension, however, under the continuous influence of climate change, the earth system has presented strong non-stationary characteristics, namely the generation mechanism of the disaster and probability distribution parameters have changed systematically and trending with time. If the stable model is continued to be used, the risk of the future extreme event is seriously underestimated or misjudged, so that the disaster prevention standard is lagged. Although some non-stationary models already exist, most of these models introduce only simple linear trends to characterize non-stationary properties, for example, assuming that the mean or standard deviation of the disaster varies linearly with time. However, the impact of climate change on disaster systems is complex and strongly nonlinear, and existing methods cannot capture this complex nonlinear and non-stationary interaction, resulting in significant bias in the estimation of the joint probability of complex extreme events. Disclosure of Invention Aiming at the defects, the invention provides a multi-disaster nonlinear and non-stationary joint probability distribution modeling method considering the influence of global climate change. Through the steps of trend analysis, periodicity analysis, non-stationary extremum analysis, non-stationary copula modeling and the like on the polar event, more accurate extreme multi-disaster non-linear non-stationary joint probability modeling is realized. The method provided by the invention provides a stronger analysis framework which is more in line with physical reality for understanding and quantifying the composite disaster risk in the climate change era, and has great theoretical value and wide application prospect. The invention provides a multi-disaster nonlinear non-stationary joint probability distribution modeling method considering climate change influence, which comprises the following steps: The first step, carrying out trend analysis on the statistical annual maximum value data, and primarily identifying time-varying characteristics; secondly, periodically analyzing the statistical annual maximum value data, identifying the periodic components hidden in the data, and obtaining dominant frequency; Thirdly, modeling nonlinear and non-stationary edge distribution of annual maximum data by adopting a non-stationary extremum distribution model, and calculating a time-varying reproduction period value; And fourthly, modeling nonlinear and non-stationary joint distribution of the annual maximum data by adopting a non-stationary copula model, and calculating a time-varying joint reproduction period contour line. Preferably, in the first step, the trend analysis is performed by sliding a time window, the time length and sliding step size of the window are firstly determined, then the mean value, standard deviation and correlation coefficient of the data in the window are calculated from the front end of the data sequence with the maximum value, the window is slid backwards according to the sliding step size, the process is repeated until the end of the data sequence is reached, and the significance level is givenThe mean value, standard deviation and correlation coefficient sequences obtained by sliding time window calculation are subjected to trend test by adopting a linear regression or M-K test method, and the test is calculatedValue of ifThe sequence is shown to have a significant trend, i.e., an immediate variable characteristic. Preferably, in the second