A Smart Aggregation Method Of Spatial-temopral Data For Natural Disaster Emergency Tasks

The rapid development of earth observation technology makes it possible to obtain large data sets with high spatial-temporal resolution and rich attribute information,and provides important support for emergency response capability of disaster management by data sharing and disaster information service.The total amount of disaster-related information and the variety of data have increased significantly,which have put forward higher requirements for data integration,processing and analysis.How to fully utilize the observation data,effectively manage the rapidly increasing multi-source heterogeneous geospatial data,automatically provide the most suitable data to the right emergency task,and meet the high precision requirement on the output of disaster reduction products,is a large challenge for improving the decision-making ability of disaster reduction.There are some problems in the current disaster data management,such as multi-source heterogeneous data are isolated from each other,data organization considers less semantic features and lack of effective correlation mechanism.The traditional disaster data retrieval method mainly relies on the simple metadata and artificial experience,which is lack of on-demand automatic data querying method.This paper proposes a smart aggregation method of spatio-temporal data for disaster emergency task,studies the emergency task requirements and transforms them into semantic constraint rules,defines the unified description and association of emergency tasks and disaster related data,proposes the multi-dimensional data feature description of semantic constraints.Based on this,we make a research on the multi-level association rules of semantic constraints,and further study the intelligent aggregation method of disaster data based on correlation analysis to achieve the active retrieval and precision discovery of disaster data set.(1)A unified description model of disaster emergency tasks and disaster-related data.This research analyzes the types,functions,requirements,and elements of the emergency task,and constructs a unified task ontology.Orienting the diversity of different data in structure,spatio-temporal description and semantic expression,this paper analyzes the temporal and spatial,granularity and attribute characteristics,expands the semantic connotation in time,space and attribute levels,and establishes the unified description of data based on ontology theory;then a multi-level semantic mapping is established between task ontology and data ontology to express the transform from task requirements to data features.(2)Association of task and data based on disaster semantics.Aiming at the problem that the diversity of disaster data type and the complexity of their semantics prevents the comprehensive correlation analysis,the "task-data" association approach based on disaster semantic constraint is proposed.In this association approach,disaster semantics is defined to describe the characteristics of disaster emergency task requirement.The method extracts the multi-dimensional semantic features of the disaster data,and establishes the association rules between task and data from spatio-temporal,attribute and preference levels.The degree of relevance of tasks and data provides method basis for automatic data aggregation.(3)Multi-level data aggregation method oriented to task requirement.In order to quickly and automatically retrieve advantage data from disaster big data for specific emergency task requirements,this paper proposes and implements a multi-level adaptive data aggregation method for automatically discovering strongly correlated data sequences that satisfy specific task requirements.In the method layer,the data filtering process based on multi-level association analysis is established,and in the implementation level,a two-layer hybrid data storage architecture is set up to support realization of aggregation method.(4)Taking the "rainstorm-flood" disaster as an example to carry out experimental application analysis.This paper analyzes the emergency tasks in the process of rainstorm and flood disaster and extracts the semantic features of tasks,then maps the task semantics and data semantics in time and space,attribute and preference level.Taking the emergency tasks such as flood simulation and disaster assessment as examples,the aggregation method automatically step by step selects the most suitable data set from massive data recommend data list to the task nodes as input.Finally,the application analysis is carried out to verify the validity of the method from the aspects of the matching degree of the recommended data and the efficiency of data retrieval.The smart aggregation method of spatial-temporal data for disaster emergency tasks firstly describes the task and data in the theoretical level systematically,then at the methodology level,a "task-data" association method based on semantic constraints is proposed and a progressive disaster information aggregation process is designed and implemented.At last,the implementation of the method is applied to the prototype of integrated management system for disaster reduction.This research helps disaster information system achieve the disaster emergency task flow automatic construction and dynamic update,automatic finding of suitable disaster data sets,which significantly improves the response and processing ability of disaster management.