CN-121999596-A - High-temperature drought composite disaster monitoring and early warning method and system

Abstract

The invention provides a high-temperature drought composite disaster monitoring and early warning method and system, and relates to the technical field of disaster monitoring and early warning, wherein the method comprises the following steps of 2, extracting a data subset from a fusion data set, constructing a characteristic analysis area according to a preset reference relation, selecting a core analysis unit as one focus of an ellipse in the characteristic analysis area, selecting an external reference unit as the other focus of the ellipse in an adjacent space range, and calculating the space morphology and dynamic change of an eccentricity quantization area of the ellipse; according to the association relation between the core analysis unit and the external reference unit along with the time change, an analysis path is generated, and the environmental parameter correction coefficient is calculated to calibrate the related elements in the fusion data set, so that a calibrated fusion data set is obtained. The method and the system realize probability prediction and dynamic early warning of the high-temperature drought composite disaster, and improve the reliability and the instantaneity of disaster early warning.

Inventors

• CHEN HUI
• LIAN YINGJIE
• ZENG JINGYU
• FENG JIANFENG
• JIANG CONG
• HE JIANQIAO
• FENG CHENG
• SHUI WEI
• TIAN JUNZHE
• FAN SHUISHENG
• TIAN JUNHUA
• WANG QIANFENG
• BAO XIAOLING
• XUE CHENGZHI

Assignees

• 硕威工程科技股份有限公司
• 福州大学
• 福建农林大学
• 福州徕斯达信息科技有限公司

Dates

Publication Date

20260508

Application Date

20260410

Claims (10)

1. 1. The high-temperature drought composite disaster monitoring and early warning method is characterized by comprising the following steps of: step 1, acquiring multi-source monitoring data in real time, and processing the multi-source monitoring data to generate a fusion data set; Step 2, extracting a data subset from the fusion data set, constructing a feature analysis area according to a preset reference relation, selecting a core analysis unit as one focus of an ellipse in the feature analysis area, selecting an external reference unit as the other focus of the ellipse in an adjacent space range, and calculating the space morphology and dynamic change of an eccentricity quantization area of the ellipse; Step 3, based on the calibrated fusion data set, constructing a time sequence dynamic analysis structure, extracting time sequence dynamic feature vectors of high temperature and drought key indexes, analyzing the change track of the key indexes, calculating the torsion angle evolving along with time, quantifying turning and fluctuation features of the trend, and fusing the time sequence dynamic feature vectors with the torsion angle features to form a comprehensive feature matrix; Step 4, inputting the comprehensive feature matrix into a pre-trained machine learning classification model, calculating the occurrence probability of the high-temperature drought composite disaster, and judging the early warning level; And step 5, issuing early warning information according to the early warning level, collecting actual disaster information fed back from outside as new training data, and periodically retraining the machine learning classification model to realize self-adaptive updating.
2. 2. The method for monitoring and early warning of a high temperature drought composite disaster according to claim 1, wherein the steps of extracting a data subset from the fused data set, constructing a feature analysis area according to a preset reference relationship, selecting a core analysis unit as one focus of an ellipse in the feature analysis area, selecting an external reference unit as the other focus of the ellipse in a neighboring spatial range, and calculating the spatial morphology and dynamic change of an eccentricity quantization area of the ellipse comprise: Selecting a target area affected by high temperature or drought based on the geospatial distribution information in the fusion data set, and determining the target area as a characteristic analysis area; In the characteristic analysis area, identifying and determining a space unit with prominent abnormal values as a core analysis unit according to the abnormal values of the meteorological and environmental elements in the fusion data set; selecting an external reference unit in the adjacent space range of the core analysis unit by taking the core analysis unit as a center according to the space relevance characteristics of the elements in the fusion data set; Taking the core analysis unit and a selected external reference unit as two focuses of the ellipse, and calculating to obtain eccentricity according to the geometric relationship between the distance between the two focuses and the length of the major axis of the ellipse; And dynamically describing the evolution process of the spatial structure morphology along time by analyzing the variation of the eccentricity on a continuous time sequence.
3. 3. The method for monitoring and early warning of a high-temperature drought composite disaster according to claim 2, wherein generating an analysis path according to the association relation between the core analysis unit and the external reference unit along with time variation, and calculating an environmental parameter correction coefficient to calibrate the relevant elements in the fusion data set, thereby obtaining a calibrated fusion data set, comprising: Calculating the association degree of the core analysis unit and the external reference unit at each time point based on the element values of the core analysis unit and the external reference unit on the multi-time sequence; Based on the association relation sequence, in a space-time coordinate system, drawing points by taking time as a horizontal axis and the association degree as a vertical axis, connecting coordinate points corresponding to all time points, and generating an analysis path for describing an association relation evolution track; Extracting curvature, direction and trend characteristics of an analysis path, and performing fusion calculation with spatial structure characteristics reflected by eccentricity to generate an environmental parameter correction coefficient; And based on the environmental parameter correction coefficient, carrying out calibration operation on the values of the temperature, precipitation and soil humidity key monitoring elements in the characteristic analysis area in the fusion data set in the corresponding time period, and generating a calibrated fusion data set.
4. 4. The method for monitoring and early warning of a high-temperature drought composite disaster according to claim 3, wherein the step of constructing a time sequence dynamic analysis structure based on the calibrated fusion data set and extracting time sequence dynamic feature vectors of high-temperature and drought key indexes comprises the following steps: Respectively selecting a key index set for representing a high-temperature event and a key index set for representing a drought event from the calibrated fusion data set; Based on the high-temperature key index and the drought key index, extracting an observation value of each key index in a preset time window from the calibrated fusion data set, and arranging the observation values according to a time sequence to construct and generate corresponding time sequence data for each key index; Based on the time series data, respectively calculating the mean value, amplitude variation, fluctuation frequency and stepwise trend slope of each piece of time series data, and extracting and generating statistical characteristics corresponding to each time series; Combining the statistics features corresponding to all the indexes in the high-temperature key index set according to a preset sequence to generate a high-temperature time sequence dynamic feature vector, and combining the statistics features corresponding to all the indexes in the drought key index set according to a preset sequence to generate a drought time sequence dynamic feature vector.
5. 5. The method for monitoring and early warning of a high temperature drought composite disaster according to claim 4, wherein analyzing the variation trace of the key index, calculating the torsion angle evolving with time, quantifying the turning and fluctuation characteristics of the trend, fusing the time sequence dynamic characteristic vector with the torsion angle characteristics to form a comprehensive characteristic matrix, comprising: Selecting a key index time sequence continuous segment with a preset time length based on index data in the time sequence dynamic feature vector, and mapping a plurality of index values corresponding to each time point of the key index time sequence continuous segment into a coordinate point in a multidimensional space to form a space track representing index change; Calculating the vector difference between coordinate points corresponding to adjacent time points based on the space track to obtain a direction vector between the adjacent time points; Based on the torsion angle sequence, calculating the maximum value of the torsion angle sequence as the maximum torsion angle, calculating the arithmetic average value of the torsion angle sequence as the average torsion angle, and calculating the times of the torsion angle exceeding a preset threshold value in unit time as the torsion frequency; And fusing the time sequence dynamic characteristic vector with the maximum torsion angle, the average torsion angle and the torsion frequency characteristic to form a comprehensive characteristic matrix.
6. 6. The method for monitoring and early warning of a high-temperature drought composite disaster according to claim 5, wherein the comprehensive feature matrix comprises original time sequence statistical features and trend turning fluctuation information.
7. 7. The method for monitoring and early warning of a high temperature drought composite disaster according to claim 6, wherein inputting the comprehensive feature matrix into a pre-trained machine learning classification model, calculating the occurrence probability of the high temperature drought composite disaster, and judging the early warning level, comprises: inputting the comprehensive feature matrix into a machine learning classification model to generate a probability value representing the occurrence possibility of the high-temperature drought composite disaster; And matching the probability value with a plurality of preset probability threshold intervals, and mapping the probability value to a corresponding early warning level based on a matching result.
8. 8. The method of claim 7, wherein the early warning level comprises at least no early warning, attention, early warning, and emergency early warning.
9. 9. The method for monitoring and early warning of a high-temperature drought composite disaster according to claim 8, wherein the method for monitoring and early warning of a high-temperature drought composite disaster is characterized by issuing early warning information according to early warning levels, collecting external feedback actual disaster information as new training data, periodically retraining a machine learning classification model, and realizing self-adaptive updating, and comprises the following steps: Based on the early warning level, generating early warning information comprising disaster intensity, influence range and defense advice, and collecting actual disaster information from meteorological sites, remote sensing monitoring platforms and on-site manual reporting in a set period after the early warning information is released; matching and correlating the actual disaster information with the corresponding comprehensive feature matrix to form a new training sample with a disaster occurrence label; And monitoring the accumulated number of the new training samples, triggering an updating process when the accumulated number reaches a preset threshold value, combining the new training samples with the original existing training samples to form an extended training data set, and retraining the machine learning classification model by using the extended training data set to realize the self-adaptive updating of the early warning.
10. 10. A high temperature drought composite disaster monitoring and early warning system implementing the method according to any one of claims 1 to 9, comprising: the fusion module is used for collecting the multi-source monitoring data in real time, processing the multi-source monitoring data and generating a fusion data set; The calibration module is used for extracting a data subset from the fusion data set, constructing a feature analysis area according to a preset reference relation, selecting a core analysis unit as one focus of an ellipse in the feature analysis area, selecting an external reference unit as the other focus of the ellipse in an adjacent space range, and calculating the space morphology and dynamic change of an eccentricity quantization area of the ellipse; The characteristic module is used for constructing a time sequence dynamic analysis structure based on the calibrated fusion data set, extracting time sequence dynamic characteristic vectors of high-temperature and drought key indexes, analyzing the change track of the key indexes, calculating the torsion angle evolving along with time, quantifying turning and fluctuation characteristics of the trend, and fusing the time sequence dynamic characteristic vectors with torsion angle characteristics to form a comprehensive characteristic matrix; The early warning module is used for inputting the comprehensive feature matrix into a pre-trained machine learning classification model, calculating the occurrence probability of the high-temperature drought composite disaster and judging the early warning grade; and the updating module is used for issuing early warning information according to the early warning grade, collecting actual disaster information fed back from outside as new training data, and periodically retraining the machine learning classification model to realize self-adaptive updating.

Description

High-temperature drought composite disaster monitoring and early warning method and system Technical Field The invention relates to the technical field of disaster monitoring and early warning, in particular to a high-temperature drought composite disaster monitoring and early warning method and system. Background In the field of high-temperature drought composite disaster monitoring and early warning, the existing technical method is mostly dependent on a single or limited data source to independently monitor high temperature or drought, for example, the high-temperature index and the drought index are calculated respectively based on the temperature and rainfall observation data of a meteorological site and are independently early warned according to the high-temperature index and the drought index, however, the high temperature and the drought tend to mutually influence and jointly evolve in the actual occurrence process to form a composite disaster, and the existing separate monitoring and early warning mode has some limitations in reflecting the characteristics of interaction and spatial collaborative evolution. For example, in the monitoring practice for a specific area, such as a hilly agricultural area, the prior art may determine a high-temperature event and a weather drought event according to the temperature and precipitation data of a limited site in the area, so that it is difficult to fully capture the heterogeneity of the spatial distribution of the high-temperature drought caused by the uneven underlying surface, such as vegetation coverage and the difference of soil humidity, and the spatial coupling form and dynamic variation of the two, for example, in the early stage of drought, the different positions of the area are affected by solar radiation, wind speed and soil conditions, the spreading of the high temperature and the development of drought are not uniformly synchronized, the spatial association form may exhibit asymmetric or directional evolution, and the existing static index analysis based on a fixed site or a regular grid mostly lacks comprehensive quantification means for the spatial form dynamics and the association with the time evolution, so that the recognition and early warning of the early signals of the occurrence and development of composite disasters may be affected. Disclosure of Invention The invention aims to solve the technical problem of providing a high-temperature drought composite disaster monitoring and early warning method and system, realizing probability prediction and dynamic early warning of the high-temperature drought composite disaster, and improving reliability and instantaneity of disaster early warning. In order to solve the technical problems, the technical scheme of the invention is as follows: in a first aspect, a method for monitoring and early warning a high-temperature drought composite disaster, the method comprising: step 1, acquiring multi-source monitoring data in real time, and processing the multi-source monitoring data to generate a fusion data set; Step 2, extracting a data subset from the fusion data set, constructing a feature analysis area according to a preset reference relation, selecting a core analysis unit as one focus of an ellipse in the feature analysis area, selecting an external reference unit as the other focus of the ellipse in an adjacent space range, and calculating the space morphology and dynamic change of an eccentricity quantization area of the ellipse; Step 3, based on the calibrated fusion data set, constructing a time sequence dynamic analysis structure, extracting time sequence dynamic feature vectors of high temperature and drought key indexes, analyzing the change track of the key indexes, calculating the torsion angle evolving along with time, quantifying turning and fluctuation features of the trend, and fusing the time sequence dynamic feature vectors with the torsion angle features to form a comprehensive feature matrix; Step 4, inputting the comprehensive feature matrix into a pre-trained machine learning classification model, calculating the occurrence probability of the high-temperature drought composite disaster, and judging the early warning level; And step 5, issuing early warning information according to the early warning level, collecting actual disaster information fed back from outside as new training data, and periodically retraining the machine learning classification model to realize self-adaptive updating. In a second aspect, a high temperature drought composite disaster monitoring and early warning system includes: the fusion module is used for collecting the multi-source monitoring data in real time, processing the multi-source monitoring data and generating a fusion data set; The calibration module is used for extracting a data subset from the fusion data set, constructing a feature analysis area according to a preset reference relation, selecting a core analysis unit as one focus of an e