Research On Visual Analysis Method Of Typhoon Intensity Change Based On Time Event Series

Typhoon is one of the most devastating and common disasters on the earth.With the intensification of climate change,the impact on human society is increasing.The research on the change of typhoon intensity has become a hot issue.At present,there is still a large error in the forecast of typhoon intensity change during the sudden change of intensity.The effective discovery and analysis of typhoon intensity change model provide support for the accurate prediction of typhoon.Visual analysis technology is one of the important means to analyze the change of typhoon intensity.The traditional visual analysis method defines the intensity changes as samples based on statistical principles,and uses bar charts and line charts to display the distribution of sample data.Only the single statistical result of typhoon intensity can be displayed,and the pattern behind the data cannot be effectively found.Typhoon intensity change patterns can be divided into two categories: one is typical typhoon patterns,most of which belong to this type of pattern and belong to the frequent pattern in typhoon intensity data set.Studying this type of pattern can find the universal law of typhoon intensity change and development,which is of great research significance.The other type is the super typhoon mode.This type of super typhoon will be strengthened rapidly in a certain period of time,which is extremely destructive.Experts in this field are more interested in this type of mode,but this type of mode belongs to the infrequent mode.The existing model visual analysis methods are mainly based on time series data for pattern mining,but directly based on typhoon intensity observation data for pattern mining will cause the explosive growth of candidate model set,unable to achieve effective discovery of typhoon patterns.Typhoon intensity changes can be abstracted as events(including Neutral event(N),Slow Intensifying event(SI),Rapid Intensifying event(RI),Weakening in this trait: The Slow Weakening event,a Rapid weakening in this trait.Several events with the same or different typhoon intensity changes were sorted by time to form a typhoon time event sequence.The method of describing the changing process of typhoon intensity based on time event series can effectively solve the problem of explosive growth of the patterns in the process of pattern mining.Therefore,how to effectively express the change of typhoon intensity as a series of time events and how to carry out frequent pattern discovery of typhoon intensity based on this series is the key issue in the current research on typhoon intensity change.At the same time,for the super typhoon model,because it is an infrequent model,the traditional frequent pattern mining method cannot achieve effective mining,so it is necessary to propose an effective visual analysis method of infrequent pattern.In view of the above problems,the specific research results of this paper are as follows:1)Proposed an unsupervised visual analysis method based on mining frequent patterns of typhoon intensity changes.Aiming at typical typhoon models,this method includes an unsupervised representation of typhoon intensity of time event sequence,which consists of three steps: event feature estimation,event sequence alignment and event sequence segmentation.This method can automatically divide the discretized typhoon intensity change sequence into different stages,in which each stage represents the intensity change rule in a period of time,which can be used to align a group of event sequence and divide it into different stages.After the phases are divided,each phase contains a set of related events that can be considered as a subset and enable frequent pattern mining within each phase.On the basis of frequent pattern mining by stage,the auxiliary display of mining results is realized through visual analysis interface.2)Based on this stage representation method of unsupervised time event series,an infrequent pattern mining visual analysis method for typhoon intensity change is proposed.This method constructs complex queries to retrieve infrequent patterns based on regular expressions and allows analysts to insert event constraints or custom pattern constraints.Combined with subsequence feature extraction and adaptive adjustment of segmentation points,segmentation can be carried out according to the structural characteristics of sequence data.At the same time,this paper also takes advantage of mining important patterns from subsequences,constructing segmentation queries from sequence segmentation in the form of constraints and using them for visual analysis.On this basis,an interactive visual analysis interface is constructed to assist the display of analysis results.A hierarchical visualization to represent multiple patterns and a timeline to reveal temporal information.The tool supports multiple interactions,such as filtering and querying,alignment,segmentation,and so on,allowing users to explore different aspects of the data by selecting different views and interactions.The experimental results show that the visual analysis method proposed in this paper can help researchers better understand the development law and trend of typhoon intensity variation data.At the same time,this method can also be applied to the visual analysis of other similar time event sequence data.