CN-122024439-A - Landslide risk early warning threshold system establishment method considering rainfall at different earlier stages

CN122024439ACN 122024439 ACN122024439 ACN 122024439ACN-122024439-A

Abstract

The invention provides a landslide risk early warning threshold system establishment method considering rainfall at different early stages, which relates to the technical field of landslide risk early warning and comprises the steps of acquiring historical landslide data, historical rainfall data and geological environment data of a research area; the method comprises the steps of selecting evaluation factors based on geological environment data, constructing a landslide vulnerability evaluation model to obtain a landslide vulnerability space partition of a research area, classifying the historical rainfall data in a rain type, determining corresponding early rainfall calculation time periods and critical rainfall calculation time periods according to rainfall concentration time periods of different rain types, constructing a rainfall threshold equation of a rainfall division type through statistical fit of the historical landslide data and the historical rainfall data, determining a rainfall early warning level according to the rainfall threshold equation, coupling the landslide vulnerability space partition with the rainfall early warning level, and constructing a danger early warning matrix to obtain a landslide risk early warning threshold system. The method can realize accurate early warning of landslide risk.

Inventors

HUANG CHENCHEN
DU JUAN
YIN KUNLONG
LIU XIJUN

Assignees

桂林理工大学

Dates

Publication Date: 20260512
Application Date: 20260122

Claims (10)

1. A landslide risk early warning threshold system establishment method considering rainfall at different earlier stages is characterized by comprising the following steps: s1, acquiring historical landslide data, historical rainfall data and geological environment data of a research area; s2, selecting an evaluation factor based on geological environment data, and constructing a landslide vulnerability evaluation model to obtain a landslide vulnerability space partition of a research area; s3, classifying the historical rainfall data based on rainfall uniformity indexes and peak position indexes, introducing a self-adaptive effective rainfall coefficient correction mechanism based on rainfall concentration degree for each type of rainfall, determining corresponding early rainfall calculation time periods and critical rainfall calculation time periods according to the rainfall concentration time periods of different rainfall types, constructing a rainfall threshold equation of the rainfall distribution type through the statistical fit of the historical landslide data and the historical rainfall data, and determining a rainfall early warning grade according to the rainfall threshold equation; and S4, coupling the landslide vulnerability space subareas with rainfall early warning grades, and constructing a danger early warning matrix to obtain a landslide risk early warning threshold system.
2. The method for establishing the landslide risk early warning threshold system considering different early-stage rainfall types according to claim 1 is characterized in that in step S2, a landslide susceptibility evaluation model adopts a machine learning model, the machine learning model is any one of a random forest model, a support vector machine model or a neural network model, a landslide sample and a non-landslide sample are trained and verified to obtain a landslide susceptibility distribution result of a research area, and a landslide susceptibility space partition is obtained by partitioning according to the landslide susceptibility distribution result.
3. The method for establishing the landslide risk early warning threshold system considering different early-stage rainfall types according to claim 2 is characterized in that a landslide vulnerability space partition is divided into a high vulnerability area, a middle vulnerability area, a low vulnerability area and a non-vulnerability area according to a landslide vulnerability probability distribution result, and the division method adopts a natural breakpoint method or a quantile method to correspondingly divide the vulnerability probability value into vulnerability areas with different grades from high to low.
4. The method for establishing a landslide risk early-warning threshold system by considering different early-stage rainfall types according to claim 1, wherein in step S3, the specific method for classifying the rainfall types is as follows: taking a Lorentz curve-based coefficient G as an index for measuring the distribution uniformity of rainfall time, wherein the calculation formula of the coefficient G is as follows: ; ; Wherein M is a Lorentzian curve; The area under the Lorentzian curve M is the range of the coefficient of Kernine G from 0 to 1; The calculation formula of the rainfall peak position coefficient r is as follows: ; In the formula, T is the total duration of a rainfall event; Combining the coefficient G of the base and the coefficient r of the peak position, dividing the rainfall process into four typical rain types of front peak type, middle peak type, rear peak type and uniform type, wherein the front peak type is And is also provided with Middle peak type is And is also provided with The rear peak is And is also provided with The uniformity is Wherein And Is the peak position coefficient threshold value, And Is the coefficient of the foundation threshold.
5. The method for establishing a landslide risk early warning threshold system by considering different early-stage rainfall patterns according to claim 4, wherein in step S3, the specific method of the adaptive effective rainfall coefficient correction mechanism based on the rainfall concentration is as follows: Introduction of an adaptive effective rainfall coefficient based on a coefficient of Kernia G The calculation formula is as follows: ; ; In the formula, The effective rainfall coefficient is used as a reference and is statistically determined through historical landslide data; Correcting intensity parameters for the coefficient of the foundation, and determining by least square fitting of historical landslide data and historical rainfall data; as a coefficient threshold, representing a transition region of concentrated rainfall and uniform rainfall; To correct the function, the influence of the rainfall concentration on the effective rainfall attenuation speed is reflected.
6. The method for establishing a landslide risk early warning threshold system considering different early-stage rainfall patterns according to claim 4, wherein in step S3, the specific calculation method of the early-stage rainfall calculation period and the critical rainfall calculation period is as follows: determining corresponding early rainfall calculation time periods and critical rainfall calculation time periods according to rainfall concentration time periods of different rain types, calculating corrected early rainfall and critical rainfall by adopting an API correction method, and correcting the corrected early rainfall The calculation formula of (2) is as follows: ; Critical rainfall The calculation formula of (2) is as follows: ; wherein j is a rainfall day number, j=0 represents the day on which landslide occurs, And The number of days of start and stop calculated for the rainfall in advance of the rain type; calculating the initial days for the critical rainfall of the rain type, wherein N is the reference days for aging attenuation; Daily rainfall on the j th day before landslide occurs is represented; Rainfall on the same day when landslide occurs; And the self-adaptive effective rainfall coefficient is calculated according to the rainfall event foundation coefficient G.
7. The method for establishing a landslide risk early warning threshold system considering different early-stage rainfall types according to claim 6, wherein in step S3, the specific method for constructing the rainfall threshold equation of the rainfall type by statistical fitting of the historical landslide data and the historical rainfall data is as follows: Carrying out probability statistics and fitting analysis on historical rainfall and landslide response data by adopting a mathematical statistics analysis method, and constructing critical rainfall threshold curves under different occurrence probability grades, wherein the general form of a rainfall threshold equation is as follows: ; wherein a and b are parameters obtained by fitting historical landslide data.
8. The method for establishing a landslide risk early-warning threshold system by considering different early-stage rainfall types according to claim 1, wherein in step S3, the method for dividing rainfall early-warning grades is as follows: According to statistical analysis of historical landslide data and historical rainfall data, determining a rainfall threshold curve corresponding to each early warning level by adopting a probability bit dividing method, dividing the rainfall early warning level into a first early warning level, a second early warning level, a third early warning level and a fourth early warning level, wherein each early warning level corresponds to different probability bit dividing values and landslide occurrence probability intervals respectively, and the early warning level rises along with the increase of the landslide occurrence probability.
9. The method for establishing a landslide risk early-warning threshold system by considering different early-stage rainfall types according to claim 1, wherein in step S4, the method for establishing the risk early-warning matrix is as follows: cross-coupling is carried out on landslide vulnerability space subareas and rainfall early warning grades, and the risk early warning matrix follows the following principles: on the basis of determining landslide susceptibility partitions, the highest warning level of the warning unit is constrained by the susceptibility level of the warning unit, wherein the non-susceptibility area does not trigger any landslide warning under any rainfall warning level condition; Under the condition of no rainfall early warning grade, each susceptibility area does not trigger landslide early warning; in the condition of the existence of the rainfall early warning level, the danger of the early warning unit and the corresponding early warning signal are gradually increased along with the rising of the rainfall early warning level in the same susceptibility level; The early warning unit is a space evaluation unit divided by a research area.
10. The method for establishing a landslide risk early warning threshold system considering different early-stage rainfall patterns according to claim 9, further comprising a dynamic optimization step based on accumulated early warning reliability scores: Maintaining an early warning reliability accumulated score S for each early warning unit, setting an initial value to 0, and updating the score according to early warning response results of each rainfall event: ; In the formula, Is the updated cumulative score; for the cumulative score before update; the score variation after a single rainfall event is determined according to the early warning response type, namely the hit time False alarm Missing reporting time When rejecting correctly Wherein 、、、 Score change values corresponding to different early warning response types are respectively obtained; And dynamically adjusting the susceptibility level of the early warning unit according to the accumulated score S: when S is lower than the first threshold The level of susceptibility is reduced by one step when S is below a second threshold At the time, the susceptibility level is reduced by one more stage, wherein ; When the missing report occurs, the susceptibility level is immediately raised by one level, and S is reset to 0; when S is higher than the third threshold At that time and the unit was downgraded during the dynamic adjustment process, the vulnerability level before downgrading is restored, S is reset to 0, where 。

Description

Landslide risk early warning threshold system establishment method considering rainfall at different earlier stages Technical Field The invention relates to the technical field of landslide risk early warning, in particular to a landslide risk early warning threshold system establishment method considering rainfall at different earlier stages. Background In recent years, the extreme climate is affected worldwide, and extreme rainfall exceeding historical meteorological hydrologic records occurs in China for many times. Early warning of landslide is space-time prediction of dangerous situations of regional landslide under extreme meteorological conditions, and the dangerous situations can be avoided. However, the existing rainfall threshold model does not consider the difference of rainfall thresholds of rainfall type geological disasters under different geological background conditions, and the early warning model needs to be improved continuously. The traditional rainfall threshold model is mainly based on the rainfall intensity-duration (I-D) or rainfall-duration (E-D) relationship, and ignores the differential influence of early rainfall types (such as space-time distribution of rainfall, occurrence time of rainfall peak values and the like) on a landslide induction mechanism. Different rainfall patterns may cause significant differences in the rainfall infiltration process and the slope stability change response, thereby affecting the probability of rainfall triggering landslide hazard. The single I-D or E-D model can not accurately reflect the rainfall threshold difference formed by the rainfall type difference, so that the early warning precision of the threshold model is insufficient. Because regional differences exist between the geological environment and the meteorological conditions of landslide, the landslide body also has great differences in the development characteristics of the material composition, the slope structure and the like. According to the geological conditions of the existing landslide of history, reasonable partition is carried out on a research area, the difference of rainfall distribution in mountain areas is emphasized, the rainfall pattern of early rainfall is reasonably partitioned, and the timeliness between landslide deformation instability and rainfall can be revealed, so that the method is important for building an area landslide disaster early warning model. In addition, the existing early warning research is carried out in a mode of judging a matrix, and the weather warning grade is obtained according to landslide vulnerability evaluation results and rainfall threshold results, so that warning information for disaster-bearing bodies cannot be provided. The Chinese patent application CN113281742A discloses an SAR landslide early warning method based on landslide deformation information and meteorological data, which comprises the steps of determining a rainfall threshold value of an effective rainfall initiating landslide by calculating a correlation coefficient between historical landslide occurrence times and time sequence rainfall before landslide occurrence, acquiring deformation data of a landslide region by utilizing a time sequence InSAR technology, and carrying out secondary early warning according to the deformation threshold value. The comprehensive early warning device combines rainfall and deformation information, and achieves comprehensive early warning of landslide. However, the method still adopts fixed rainfall coefficient during rainfall processing, cannot distinguish the differential influence of different rainfall time distribution modes on landslide triggering, and lacks the fine consideration of early rainfall aging attenuation characteristics. Disclosure of Invention In view of the above, the invention provides a landslide risk early warning threshold system establishment method considering different early-stage rainfall types, which is characterized in that a rainfall type classification is carried out on a rainfall process, a self-adaptive effective rainfall coefficient correction mechanism and a rainfall calculation method based on aging attenuation are introduced for different rainfall types, a rainfall threshold equation of a rainfall distribution type is established, and a landslide vulnerability space partition is coupled with a rainfall early warning level, so that a risk early warning matrix is established, and accurate early warning on landslide risks is realized. The technical scheme of the invention is realized as follows: The invention provides a landslide risk early warning threshold system establishment method considering rainfall at different earlier stages, which comprises the following steps: s1, acquiring historical landslide data, historical rainfall data and geological environment data of a research area; s2, selecting an evaluation factor based on geological environment data, and constructing a landslide vulnerability evaluati