A cloud computing-based tsunami disaster risk assessment method and system and a medium

Abstract

The present application relates to the technical field of data processing, and more particularly to a tsunami disaster risk assessment method and system based on cloud computing and a medium. The method comprises the following steps: accessing and converging data of an earthquake monitoring system, a satellite remote sensing system, a tide level monitoring device and a historical tsunami event library to obtain a multi-source heterogeneous tsunami original data set; performing format standardization, time synchronization and semantic annotation processing on the multi-source heterogeneous tsunami original data set to obtain a unified data model; establishing and registering a data service interface based on the unified data model to obtain a data service set; and calling the data service set to obtain tsunami propagation simulation input data. Through multi-source data standardization and semantic fusion, cloud computing parallel processing, multi-dimensional index analysis and dynamic visual display, the present application realizes efficient, accurate, real-time and operable comprehensive management of tsunami disaster risk assessment.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a cloud computing-based method, system, and medium for tsunami disaster risk assessment. Background Technology

[0002] Early risk assessments primarily relied on historical tsunami event records and tide monitoring data, using empirical models to estimate tsunami energy and impact range. However, these methods are limited by sample size and regional applicability, making it difficult to accurately reflect complex earthquake mechanisms and real-time propagation processes. Subsequently, with the development of computer simulation technology, researchers introduced numerical methods such as finite difference and finite volume to establish tsunami propagation models based on seismic parameters and topographic data, enabling dynamic prediction of wave height, arrival time, and impact area. However, existing technologies still have significant shortcomings: firstly, the inconsistent formats and semantics of multi-source data make it difficult to integrate and share data across systems; secondly, traditional numerical simulations rely on single-node or cluster computing, making it difficult to achieve large-scale parallel computing and real-time response in sudden events; and thirdly, assessment results are mostly presented in static charts, lacking interactive display and dynamic updating capabilities, making it difficult to support multi-departmental collaborative decision-making. Summary of the Invention

[0003] Therefore, it is necessary for the present invention to provide a cloud computing-based tsunami disaster risk assessment method, system, and medium to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a cloud computing-based tsunami disaster risk assessment method includes the following steps:

[0005] Step S1: Access and aggregate data from earthquake monitoring systems, satellite remote sensing systems, tide level monitoring devices, and historical tsunami event databases to obtain a multi-source heterogeneous tsunami raw dataset; perform format standardization, time synchronization, and semantic annotation on the multi-source heterogeneous tsunami raw dataset to obtain a unified data model;

[0006] Step S2: Establish and register the data service interface based on the unified data model to obtain the data service set; call the data service set to obtain the tsunami propagation simulation input data;

[0007] Step S3: Divide and containerize the input data of the tsunami propagation simulation to obtain a set of multi-node parallel computing tasks; dynamically schedule and solve the set of multi-node parallel computing tasks in parallel to obtain partitioned computing results.

[0008] Step S4: Synchronize and fuse the partition calculation results to obtain the global tsunami propagation result dataset;

[0009] Step S5: Perform statistical analysis and index extraction on the global tsunami propagation result dataset to obtain a tsunami disaster risk assessment index set;

[0010] Step S6: Render and dynamically display the tsunami disaster risk assessment index set to obtain interactive risk distribution results.

[0011] Preferably, the present invention also provides a cloud computing-based tsunami disaster risk assessment system for executing the above-described cloud computing-based tsunami disaster risk assessment method, wherein the cloud computing-based tsunami disaster risk assessment system includes:

[0012] The multi-source tsunami data access and preprocessing module is used to access and aggregate data from earthquake monitoring systems, satellite remote sensing systems, tide level monitoring devices, and historical tsunami event databases to obtain a multi-source heterogeneous tsunami raw dataset; the multi-source heterogeneous tsunami raw dataset is then processed for format standardization, time synchronization, and semantic annotation to obtain a unified data model;

[0013] The unified data service construction and invocation module is used to establish and register data service interfaces based on a unified data model to obtain a data service set; and to invoke the data service set to obtain tsunami propagation simulation input data.

[0014] The cloud-native parallel computing and task scheduling module is used to partition and containerize the input data of the tsunami propagation simulation to obtain a multi-node parallel computing task set; and to dynamically schedule and solve the multi-node parallel computing task set in parallel to obtain partitioned computing results.

[0015] The partition result synchronization and data fusion module is used to synchronize and fuse the partition calculation results to obtain a global tsunami propagation result dataset.

[0016] The risk indicator analysis and assessment module is used to perform statistical analysis and indicator extraction on the global tsunami propagation result dataset to obtain a set of tsunami disaster risk assessment indicators.

[0017] The risk visualization rendering and interactive display module is used to render and dynamically display the tsunami disaster risk assessment index set, and obtain interactive risk distribution results.

[0018] Preferably, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the cloud computing-based tsunami disaster risk assessment method described above.

[0019] This invention achieves efficient integration and standardization of earthquake monitoring, satellite remote sensing, tide monitoring, and historical tsunami event data by constructing a unified multi-source data access and semantic fusion mechanism. This enables heterogeneous data to be shared and accessed on the same platform, significantly improving data utilization efficiency and consistency. Utilizing multi-node parallel computing and dynamic task scheduling technology in a cloud computing environment, it achieves high-performance processing of large-scale tsunami propagation calculation tasks, rapidly generating high-precision wave height, propagation speed, and impact range information during sudden earthquake events, solving the problem of slow response in traditional single-node or cluster computing. Through synchronous fusion of partitioned calculation results and global data consistency processing, it ensures the spatial continuity and temporal consistency of simulation output results, providing an accurate and reliable data foundation for subsequent risk assessment. By leveraging multi-dimensional indicator statistical analysis and searchable data structures, it enhances the refinement and operability of tsunami risk assessment indicators, enabling different departments to quickly query and analyze based on the same data source. Simultaneously, through interactive visualization rendering and dynamic display, it supports real-time observation, interactive exploration, and decision support of risk distribution, providing intuitive and flexible data support for emergency management and multi-departmental collaborative decision-making, thereby comprehensively improving the accuracy, real-time performance, and operability of tsunami disaster risk assessment. Attached Figure Description

[0020] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0021] Figure 1 This is a schematic diagram of the steps of a cloud computing-based tsunami disaster risk assessment method according to the present invention.

[0022] Figure 2 This is a schematic diagram of the system modules of a cloud computing-based tsunami disaster risk assessment system according to the present invention;

[0023] Figure 3 This is a schematic diagram of cloud-native parallel computing and task scheduling in this invention. Detailed Implementation

[0024] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0025] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0026] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0027] To achieve the above objectives, please refer to Figures 1 to 3 This invention provides a cloud computing-based tsunami disaster risk assessment method, the method comprising the following steps:

[0028] Step S1: Access and aggregate data from earthquake monitoring systems, satellite remote sensing systems, tide level monitoring devices, and historical tsunami event databases to obtain a multi-source heterogeneous tsunami raw dataset; perform format standardization, time synchronization, and semantic annotation on the multi-source heterogeneous tsunami raw dataset to obtain a unified data model;

[0029] Step S2: Establish and register the data service interface based on the unified data model to obtain the data service set; call the data service set to obtain the tsunami propagation simulation input data;

[0030] Step S3: Divide and containerize the input data of the tsunami propagation simulation to obtain a set of multi-node parallel computing tasks; dynamically schedule and solve the set of multi-node parallel computing tasks in parallel to obtain partitioned computing results.

[0031] Step S4: Synchronize and fuse the partition calculation results to obtain the global tsunami propagation result dataset;

[0032] Step S5: Perform statistical analysis and index extraction on the global tsunami propagation result dataset to obtain a tsunami disaster risk assessment index set;

[0033] Step S6: Render and dynamically display the tsunami disaster risk assessment index set to obtain interactive risk distribution results.

[0034] Preferably, step S1 includes the following steps:

[0035] Step S11: Collect and access data from the earthquake monitoring system, satellite remote sensing system, tide level monitoring device, and historical tsunami event database to obtain the raw multi-source data stream;

[0036] Step S12: Perform data parsing and protocol conversion on the original multi-source data stream to obtain standardized data packets;

[0037] Step S13: Perform time alignment and geospatial registration on the standardized data packets to obtain a spatiotemporally aligned dataset;

[0038] Step S14: Based on the spatiotemporal aligned dataset, perform unified semantic mapping and label annotation on the attribute fields of each monitoring source to obtain a semantic fusion dataset;

[0039] Step S15: Extract metadata and construct a data index from the semantic fusion dataset to obtain a searchable multi-source heterogeneous tsunami original dataset;

[0040] Step S16: Perform model formatting and structured encapsulation on the searchable multi-source heterogeneous tsunami raw dataset to obtain a unified data model.

[0041] In this embodiment of the invention, firstly, in step S11, data generated by the earthquake monitoring system, satellite remote sensing system, tide monitoring device, and historical tsunami event database are synchronously acquired and network-accessed through a seismic waveform acquisition terminal deployed at the earthquake monitoring center, a remote sensing satellite ground receiving station equipped with C-band and L-band sensors, an automatic tide level recorder along the coast, and a historical tsunami event database interface module. The acquisition frequency is set to 10 times per second, forming a continuous raw multi-source data stream. Subsequently, in step S12, data is acquired through Apache... NiFi's data stream processing engine performs protocol parsing and format conversion on the raw multi-source data streams, converting seismic waveform data from SEED format to JSON structure, remote sensing image data from GeoTIFF format to HDF5 format, and tide time series data from CSV format to Parquet format, all uniformly encoded using the UTF-8 character set to obtain standardized data packets. In step S13, the Hadoop distributed time synchronization algorithm is used to compare and correct the timestamps of the standardized data packets, achieving time alignment based on UTC time. Simultaneously, geospatial registration of spatial location parameters is performed based on the WGS-84 geographic coordinate system, forming a spatiotemporally aligned dataset. In step S14, a semantic mapping algorithm based on an ontology dictionary is used to uniformly semantically label the field attributes in the spatiotemporally aligned dataset, including earthquake magnitude and focal depth. The core indicators such as intensity, seabed topography, and wave height are labeled and stored using an RDF triplet structure to generate a semantic fusion dataset. Next, in step S15, metadata extraction and multi-level index construction are performed on the semantic fusion dataset based on the Elasticsearch distributed index engine. The index fields are divided into three categories: time, space, and physical attributes. An inverted index structure is established and a search keyword dictionary is generated to form a searchable multi-source heterogeneous tsunami raw dataset. Finally, in step S16, Spark structured encapsulation tasks are scheduled using a cloud-based Kubernetes cluster to perform structured encapsulation operations on the searchable multi-source heterogeneous tsunami raw dataset. The field hierarchy is reorganized according to data type and time series order, and the Avro binary structured storage format is uniformly adopted. Field validation and consistency correction are performed to finally generate a unified data model.

[0042] Preferably, step S2, which involves establishing and registering the data service interface based on a unified data model, includes:

[0043] Based on a unified data model, attribute extraction and interface definition are performed on various data entities to obtain a data service description file.

[0044] The data service description file is encapsulated and registered to obtain a set of data service interfaces.

[0045] In this embodiment of the invention, firstly, based on the unified data model obtained in step S1, attribute extraction and interface definition operations are performed on various data entities. An attribute parsing script written in Python is run in a Spark cluster to perform field-level parsing and data type identification on key fields such as "magnitude," "focal depth," "wave height," "propagation speed," "tidal change rate," and "remote sensing reflectivity" in the data entities. Numerical fields are limited to double-precision floating-point types, time fields are limited to UTC-formatted timestamp types, and geospatial fields are limited to latitude and longitude dual-value structures. Subsequently, an attribute description file is generated for each type of data entity using the Apache Avro schema definition mechanism, and a data service description file is established based on this file. This file defines field names, field types, calling methods, and interface permissions. The calling methods adopt a RESTful architecture design, and permission management uses OAuth2 authentication. After completing attribute extraction and interface definition, Spring... The OpenFeign component in the cloud microservice framework encapsulates and registers data service description files, turning each interface into an independent microservice unit. Each unit corresponds to an access service for a specific data entity, and the service call path uses a unified naming convention: " / api / v1 / tsunami / {entity}". For example, the earthquake event entity interface is named " / api / v1 / tsunami / earthquake", and the tide monitoring entity interface is named " / api / v1 / tsunami / tide". Subsequently, all interfaces are uniformly registered with the cloud service registry Consul, which runs on the master node of the Kubernetes cluster via HTTP. Port 8500 implements interface heartbeat detection and status synchronization; after interface registration, a data service interface set is generated and stored in the distributed object storage system MinIO in JSON format. Each interface record includes interface identifier, access path, input and output parameter types, call frequency limit, and update timestamp fields; after registration, the system assigns access policies and call limits to each interface in the cloud configuration center. The call frequency limit is set to no more than 200 times per minute to ensure that each interface can respond stably and provide tsunami propagation simulation input data in a unified format when calling the data service set in the future.

[0046] Preferably, the invocation of the data service set in step S2 includes:

[0047] The set of callable data service interfaces is orchestrated and access authorized to obtain a data service set.

[0048] Parameter filtering and call scheduling are performed on the data service set to obtain the target data request task set;

[0049] Real-time data retrieval and cache preprocessing are performed on the target data request task set to obtain tsunami propagation simulation input data.

[0050] In this embodiment of the invention, firstly, service orchestration and access authorization are implemented on the set of callable data service interfaces. Apache Airflow is used to orchestrate the interface call sequence in a directed acyclic graph (DAG) manner. Kong is used as the call gateway, and Keycloak's OAuth2-based Bearer Token mechanism is used for authentication and authorization. Service registration information is provided by Consul. The orchestration task runs as a Pod in the Kubernetes namespace. The orchestration rules are limited to: a maximum of 20 concurrent tasks in the same DAG, a single task timeout of 120 seconds, and the interface call concurrency is controlled by Kong's upstream rate limiting parameters to no more than 5 requests per second. After obtaining the orchestrated and authorized data service set, parameter filtering and call scheduling are performed. Call scheduling is executed by Airflow's task scheduler combined with a parameter parser. The parameter parser uses fields in the unified data model as the standard, according to the time range field (start time and end time, represented by UTC timestamps), the spatial range field (minimum longitude, maximum longitude, minimum latitude, maximum latitude, represented by WGS-84 coordinate system), and the attribute field (magnitude threshold). 4.0, Threshold for rate of change of tide level Requests meeting the criteria are filtered using three rules (0.05 m / min). The filtering results form a target data request task set, where each task includes at least: request ID, target interface path, parameter set, expected return format (Avro binary), concurrency limit, and maximum retries. Call scheduling is performed according to a priority queue, with priority determined by time urgency and geographical coverage. Low-priority tasks wait in the queue for no more than 300 seconds before being downgraded and merged. Real-time data retrieval and cache preprocessing are then implemented on the target data request task set. Retrieval is divided into two channels: a streaming data channel using Kafka consumer groups, with parallelism limited to no more than 10 partition consumption threads per topic. Each consumer retrieves 500 records at a time and writes them to a Redis cache queue. Redis keys are named in the format "tsunami:input:{interface identifier}:{UTC minute time window}" and have a TTL of 300 seconds. A point-to-point data query channel uses RESTful synchronous retrieval, initiated by an HTTP client using a Keep-Alive connection reuse method. The request header includes a Bearer header. The token and call ID are used for request failures. Retrying follows an exponential backoff strategy with an initial backoff delay of 500 milliseconds and a maximum of 5 retries. The final response undergoes format and pattern validation immediately upon receipt, using the pattern definition from the Avro pattern registry. Records failing validation are written to an error log with an error code for subsequent auditing. During the cache preprocessing stage, time window alignment is performed on the retrieved data using a sliding window alignment algorithm. The time alignment precision is 1 second, the window width is ±1 second, and the window uses a nearest-neighbor padding strategy. Quadtree tile index mapping is performed on spatial point data for subsequent parallel reading. Numeric fields undergo type conversion and unit unification, are forced to double-precision floating-point types, and null values ​​are padded with IEEE-754. NaN identifiers are used to record the source interface location; all preprocessed and validated data is written to the MinIO object storage in the Avro mode of a unified data model. The storage path adopts the format " / input / {interface identifier} / {YYYYMMDD} / {HHMMSS}-{request ID}.avro". The write operation adopts batch write, with a maximum of 10,000 records in a single batch file. After writing, the MD5 checksum is calculated and written to the object metadata. The completed file is submitted to the task status database (PostgreSQL). The record fields include file path, generation time, number of records, MD5 checksum, and processing time. This batch-written file serves as the input data for the tsunami propagation simulation.

[0051] Preferably, step S3 includes the following steps:

[0052] Step S31: Perform spatial grid division and computational region identification on the tsunami propagation simulation input data to obtain a partitioned dataset;

[0053] Step S32: Perform granular segmentation and dependency analysis on the computation task based on the partitioned dataset to obtain the distributed task description file;

[0054] Step S33: Containerize the distributed task description file, distribute it across multiple nodes, and dynamically schedule it to obtain the partitioned computation results.

[0055] In this embodiment of the invention, firstly, in step S31, the tsunami propagation simulation input data generated in step S2 is spatially gridded and the calculation area is identified. A latitude-longitude-based regular grid algorithm is used to divide the study area into grid cells with a side length of 0.5 kilometers. Each grid cell is assigned a unique ID and its latitude-longitude range and administrative region information are recorded, generating a partitioned dataset. Simultaneously, the initial wave height, tide level, and source parameters of each grid cell are bound to a timestamp. In step S32, based on the partitioned dataset, the calculation tasks are granularly segmented, defining the calculation task of each grid cell as an independent task unit. The hydrodynamic propagation relationship and boundary condition propagation between grid cells are analyzed using a task dependency graph. In step S33, the distributed task description files are generated sequentially. These files contain the task ID, input data path, a list of dependent task IDs, computation priority, and estimated computation time. The distributed task description files are then containerized and image-built using Docker. Each task unit is represented by an image containing the tsunami propagation numerical calculation program, dependent libraries, and the task description file. The image tag is named "tsunami_task_{task ID}", and the image size is kept under 500MB. The deployable task image set is then distributed to each node of the Kubernetes cluster, and containers are instantiated on the nodes, with the container resource limit set to CPU. With 4 cores and 16GB of memory, each node can run no more than 8 containers simultaneously, generating a multi-node parallel computing task set. Each container mounts its input and output data paths through a shared volume. Resource monitoring and dynamic scheduling are performed on the multi-node parallel computing task set. Prometheus is used to collect CPU, memory, and network usage for each container. Combined with the Kubernetes scheduler, task migration or delayed scheduling operations are performed on nodes with excessive load, generating a scheduling optimization parameter set. Each parameter record includes the task ID, target node, allocated CPU cores, memory quota, and priority adjustment value. Based on the scheduling optimization parameter set, the parallel computing task set is solved in parallel and the results are cached. The Spark distributed computing engine is used to run a tsunami propagation numerical calculation program within the container, with a fixed calculation time step of 1 second. Wave height and tide level calculations are retained to two decimal places. Data calculations for each time step are written to a Redis cache queue. Simultaneously, calculation result files in Avro format are stored in MinIO object storage by container ID, task ID, and time step, with MD5 checksums recorded. Finally, partitioned calculation results are obtained, and the partitioned calculation results form a full dataset according to grid ID and time order.

[0056] Of particular importance, step S32 includes:

[0057] Step S321: Based on the partitioned dataset, evaluate the workload and quantify the computational complexity of each computational sub-region to obtain the task granularity partitioning strategy;

[0058] Step S322: Based on the task granularity partitioning strategy, perform operator decomposition and dependency analysis on the tsunami propagation simulation process to obtain the task dependency graph;

[0059] Step S323: Based on the task dependency graph, perform input-output mapping and communication overhead evaluation on each computing task node to obtain the node communication configuration set;

[0060] Step S324: Structure and encapsulate the task granularity partitioning strategy, task dependency graph, and node communication configuration set to obtain a distributed task description file.

[0061] In this embodiment of the invention, firstly, in step S321, the workload and computational complexity of each computational sub-region are evaluated and quantified based on the partitioned dataset generated in step S31. A granular analysis method is used to estimate the computational workload of each grid cell based on wave height, tide level, propagation speed, and time step length. The computational workload is measured in floating-point operations per second (FLOPS), and a task granularity partitioning strategy is generated. Each strategy record includes the sub-region ID, estimated computational workload, time step count, number of grid cells, and number of dependency boundaries. Secondly, in step S322, the tsunami propagation simulation process is decomposed into operators and its dependencies are resolved based on the task granularity partitioning strategy. A DAG graph structure resolution algorithm is used to split each computational task into independent operator nodes, resolving the dependencies between operators, including predecessor operators, successor operators, and data transfer fields, generating a task dependency graph. Each edge in the graph records the source operator ID, target operator ID, dependency fields, and data size. In step S323... In step S323, input-output mapping and communication overhead assessment are performed on each computation task node based on the task dependency graph. The data transmission latency of each dependency edge is calculated using network bandwidth and node memory capacity data, generating a node communication configuration set. Each record contains source node ID, target node ID, transmission data field, data size, estimated transmission time, and priority, which are used for subsequent scheduling optimization. In step S324, the task granularity partitioning strategy, task dependency graph, and node communication configuration set are structurally integrated and parameters are encapsulated. The three types of data are uniformly encapsulated using JSON format, with fields strictly limited as follows: sub-region ID is a string, computation amount is an integer, time step count is an integer, grid cell count is an integer, dependency boundary count is an integer, operator ID is a string, data field name is a string, data size is a double-precision floating-point number, estimated transmission time is a floating-point number, and priority is an integer. The encapsulation result is the distributed task description file.

[0062] Preferably, step S4 includes the following steps:

[0063] Step S41: Perform spatiotemporal consistency verification and boundary matching on the partition calculation results to obtain a spatial stitching reference set;

[0064] Step S42: Perform numerical fusion and global coordinate correction on the spatial stitching reference set to obtain a preliminary global result dataset;

[0065] Step S43: Perform multi-source consistency integration on the preliminary global result dataset to obtain the global tsunami propagation result dataset.

[0066] In this embodiment of the invention, firstly, in step S41, the partition calculation results generated in step S3 are compared with timestamps and verified for versions. A distributed time series comparison algorithm is used to verify the timestamps of each partition data file using UTC time as the standard to ensure continuity and consistency of order. Version consistency is verified by comparing file hash values, generating a partition result verification dataset. Each record contains a partition ID, timestamp, version number, and verification status. Based on the partition result verification dataset, overlap detection and boundary matching are performed on the spatial boundary data output by each computing node. The quadtree spatial indexing method is used to match and compare adjacent partition boundaries, detect overlapping areas, and generate a spatial stitching reference set, recording the start and end latitude and longitude of each boundary unit, adjacent partition IDs, and matching error values. Subsequently, in step S42, the spatial stitching reference set is interpolated and numerically smoothed. A bilinear interpolation algorithm is used to interpolate and fill the wave height and tide level data in the boundary missing or overlapping areas, and a three-point moving average filter is used to further smooth the data. After interpolation, the data is smoothed to generate a regional fusion dataset. Each record contains a grid ID, latitude and longitude range, wave height, tide level, and processing timestamp. In step S44, the regional fusion dataset undergoes global coordinate unification and precision correction. Based on the WGS-84 coordinate system, the coordinates of all grid cells are standardized. At the same time, the precision of wave height, tide level, and other values ​​is corrected to two decimal places using a floating-point rounding strategy, forming a preliminary global result dataset, and the correction parameters are recorded. In step S43, the preliminary global result dataset undergoes multi-source consistency comparison and metadata integration. A parallel hash comparison algorithm is used to check the field consistency of corresponding records from earthquake, tide level, and remote sensing data sources. The latitude and longitude, timestamp, wave height, tide level, and source parameter fields are integrated into a unified metadata structure, and a global tsunami propagation result dataset is generated. This global dataset is sorted by time step and grid ID and stored in MinIO object storage. Each record contains the source data ID, processing status, and generation time.

[0067] Preferably, step S5 includes the following steps:

[0068] Step S51: Perform semantic parsing, metadata extraction, and index encoding on the global tsunami propagation result dataset to obtain a preliminary data index table;

[0069] Step S52: Optimize the weights and configure the retrieval rules for the preliminary data index table to obtain a searchable multi-source heterogeneous tsunami raw dataset.

[0070] In this embodiment of the invention, firstly, in step S51, the global tsunami propagation result dataset generated in step S4 is subjected to semantic parsing and feature annotation. Using a semantic mapping method based on an ontology dictionary, the fields in the dataset are uniformly semantically labeled. Core fields include "magnitude," "focal depth," "wave height," "tidal change rate," and "propagation speed." Each field is labeled using an RDF triple structure, and the annotation information is written into an HBase distributed table to generate a semantic fusion dataset. Each record contains a grid ID, timestamp, latitude and longitude, physical attributes, and semantic tags. Subsequently, in step S52, metadata extraction and field mapping operations are performed on the semantic fusion dataset using Apache. The Spark parallel computing engine extracts the time, spatial, and numerical fields of each record and maps them to a unified metadata structure, generating a multi-dimensional metadata collection. Each record includes the field name, field type, unit, data source, and update time. Based on this multi-dimensional metadata collection, data attributes, timelines, and spatial coordinates are indexed and encoded. Three types of indexes are built using the Elasticsearch distributed indexing engine: the time index uses UTC timestamps as keys, the spatial index uses a quadtree structure to divide latitude and longitude regions, and the attribute index is segmented and encoded according to magnitude, tide level, and wave height values, generating a preliminary data index table. Each index record includes a grid ID, timestamp, spatial range, attribute field, and index identifier. The preliminary data index table is then weighted and retrieval rules are configured. An inverted index method is used to assign weight values ​​to attribute fields, and the time and spatial indexes are arranged in strict order. Boolean and range retrieval rules are constructed, with a retrieval concurrency of 50 times per second and a cache hit rate controlled above 80%. Finally, a searchable multi-source heterogeneous tsunami raw dataset is generated, where each record includes an index ID, field value, weight, and metadata identifier.

[0071] Of particular importance, step S54 includes:

[0072] Step S541: Based on the multi-dimensional metadata set, perform retrieval frequency statistics and correlation analysis on the index fields to obtain the field weight coefficient set;

[0073] Step S542: Optimize the index structure and prioritize the preliminary data index table based on the field weight coefficient set to obtain the optimized index table;

[0074] Step S543: Define rules for the combination of query conditions and matching logic based on the preset retrieval scenario to obtain a retrieval rule set;

[0075] Step S544: Associate and bind the optimized index table with the retrieval rule set and integrate the strategies to obtain a searchable multi-source heterogeneous tsunami raw dataset.

[0076] In this embodiment of the invention, firstly, in step S541, based on the multi-dimensional metadata set generated in step S53, the index fields are subjected to retrieval frequency statistics and correlation analysis. The Apache Spark distributed computing engine is used to statistically analyze the access frequency of each field over the past 30 days. The statistical fields include time, space, magnitude, tide, and wave height. Simultaneously, the joint access frequency between fields is calculated through correlation analysis, generating a set of field weight coefficients. Each record includes the field name, access frequency, joint access count, correlation index, and calculation timestamp. Subsequently, in step S542, based on the set of field weight coefficients, the preliminary data index table generated in step S53 is optimized in terms of index structure and priority sorting. The Elasticsearch index optimization algorithm is used to create a primary index for high-weight fields and an auxiliary index for low-weight fields. Simultaneously, a composite sorting rule is set according to ascending order of the time field, latitude and longitude order of the space field, and attribute field weight values ​​to generate an optimized index table. Each index record includes the index ID, field value, field weight, index type, and ranking. Sequence number; In step S543, rules are defined for query condition combination and matching logic based on preset retrieval scenarios. Retrieval scenarios include time range query, spatial region query, attribute threshold query, and multi-condition compound query. Query condition combination rules are defined in Boolean logic, and condition priority and field weight are set to generate a retrieval rule set. Each rule record includes condition type, involved fields, logical operators, priority, and matching strategy. Subsequently, in step S544, the optimized index table is associated and bound with the retrieval rule set and strategy is integrated. The index mapper is used to map each rule to the fields of the optimized index table to form a retrieval strategy file, which is written to Elasticsearch or Redis to support high-concurrency queries. Finally, a searchable multi-source heterogeneous tsunami raw dataset is generated. Each record includes index ID, field value, field weight, rule ID, and strategy identifier.

[0077] Preferably, step S6 includes the following steps:

[0078] Step S61: Perform model field filtering and structural reorganization on the searchable multi-source heterogeneous tsunami raw dataset to obtain the model mapping dataset;

[0079] Step S62: Based on the model mapping dataset, perform unified encoding and structural association on the data semantic layer and logical layer to obtain a standardized data mapping table;

[0080] Step S63: Perform model formatting and structured encapsulation on the standardized data mapping table to obtain a unified data model;

[0081] Step S64: Perform cloud-based verification and service registration on the unified data model to obtain a reusable model service interface.

[0082] In this embodiment of the invention, firstly, in step S61, the searchable multi-source heterogeneous tsunami raw dataset generated in step S5 is subjected to model field filtering and structural reorganization. The Spark SQL distributed query engine is used to filter the time field, spatial field, physical attribute field, and semantic label field in the dataset, and grouping and sorting them by grid ID and timestamp. The field structure is reorganized in the order of "time—space—attribute—semantic label" to generate a model mapping dataset. Each record contains grid ID, UTC timestamp, latitude and longitude range, magnitude, wave height, tide level, propagation speed, and semantic label. In step S62, based on the model mapping dataset, the semantic layer and logical layer of the data are uniformly encoded and structurally associated. The RDF triplet encoding method is used to generate a unique URI for each record. Semantic layer fields and logical layer data fields are associated through URIs, establishing cross-field reference relationships and forming a standardized data mapping table. Records in the table contain URI identifiers, original field names, logical field names, and data classes. The data model is generated by formatting and structuring the standardized data mapping table in step S63. Field serialization is performed using Avro binary format, and semantic layer fields are placed before logical layer fields according to the hierarchical structure. Field types are strictly limited to: time fields are UTC timestamps, spatial fields are double-precision floating-point latitude and longitude, and numerical fields are double-precision floating-point types. During the encapsulation process, field integrity verification and data consistency checks are performed, and the verification information is written to the object metadata. In step S64, the unified data model is verified and registered in the cloud. Cloud verification uses a containerized verification task running within a Kubernetes cluster. Verification includes field integrity, index consistency, and interface response testing. After successful verification, the unified data model is registered as a reusable model service interface using the Spring Cloud microservice framework. The interface path uses the unified naming rule " / api / v1 / tsunami / model / {grid ID}", and interface access uses OAuth2 Bearer Token authentication. Service information is stored in the Consul service registry, recording the interface identifier, URI path, field list, data type, and update time.

[0083] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is not limited by the foregoing description. Thus, all changes falling within the meaning and scope of the equivalents of the application are intended to be included within the scope of the invention.

[0084] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A cloud computing-based method for tsunami disaster risk assessment, characterized in that, Includes the following steps: Step S1: Access and aggregate data from earthquake monitoring systems, satellite remote sensing systems, tide level monitoring devices, and historical tsunami event databases to obtain a multi-source heterogeneous tsunami raw dataset; perform format standardization, time synchronization, and semantic annotation on the multi-source heterogeneous tsunami raw dataset to obtain a unified data model; Step S2: Establish and register data service interfaces based on a unified data model to obtain a data service set; The data service set is invoked to obtain the tsunami propagation simulation input data; wherein, the invocation of the data service set in step S2 includes: The set of callable data service interfaces is orchestrated and access authorized to obtain a data service set. Parameter filtering and call scheduling are performed on the data service set to obtain the target data request task set; Real-time data retrieval and cache preprocessing are performed on the target data request task set to obtain tsunami propagation simulation input data; Step S3: The input data from the tsunami propagation simulation is partitioned and containerized for deployment to obtain a multi-node parallel computing task set; this multi-node parallel computing task set is dynamically scheduled and solved in parallel to obtain partitioned computing results; Step S3 includes the following steps: Step S31: Perform spatial grid division and computational region identification on the tsunami propagation simulation input data to obtain a partitioned dataset; Step S32: Perform granular segmentation and dependency analysis on the computation task based on the partitioned dataset to obtain the distributed task description file; Step S33: Containerize the distributed task description file, distribute it across multiple nodes, and dynamically schedule it to obtain the partitioned computation results; Step S4: Synchronize and fuse the partitioned calculation results to obtain the global tsunami propagation result dataset; Step S4 includes the following steps: Step S41: Perform spatiotemporal consistency verification and boundary matching on the partition calculation results to obtain a spatial stitching reference set; Step S42: Perform numerical fusion and global coordinate correction on the spatial stitching reference set to obtain a preliminary global result dataset; Step S43: Perform multi-source consistency integration on the preliminary global result dataset to obtain the global tsunami propagation result dataset; Step S5: Perform statistical analysis and index extraction on the global tsunami propagation result dataset to obtain a tsunami disaster risk assessment index set; Step S6: Render and dynamically display the tsunami disaster risk assessment index set to obtain interactive risk distribution results.

2. The cloud computing-based tsunami disaster risk assessment method according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Collect and access data from the earthquake monitoring system, satellite remote sensing system, tide level monitoring device, and historical tsunami event database to obtain the raw multi-source data stream; Step S12: Perform data parsing and protocol conversion on the original multi-source data stream to obtain standardized data packets; Step S13: Perform time alignment and geospatial registration on the standardized data packets to obtain a spatiotemporally aligned dataset; Step S14: Based on the spatiotemporal aligned dataset, perform unified semantic mapping and label annotation on the attribute fields of each monitoring source to obtain a semantic fusion dataset; Step S15: Extract metadata and construct a data index from the semantic fusion dataset to obtain a searchable multi-source heterogeneous tsunami original dataset; Step S16: Perform model formatting and structured encapsulation on the searchable multi-source heterogeneous tsunami raw dataset to obtain a unified data model.

3. The cloud computing-based tsunami disaster risk assessment method according to claim 1, characterized in that, Step S2, which involves establishing and registering the data service interface based on the unified data model, includes: Based on a unified data model, attribute extraction and interface definition are performed on various data entities to obtain a data service description file. The data service description file is encapsulated and registered to obtain a set of data service interfaces.

4. The cloud computing-based tsunami disaster risk assessment method according to claim 1, characterized in that, Step S5 includes the following steps: Step S51: Perform semantic parsing, metadata extraction, and index encoding on the global tsunami propagation result dataset to obtain a preliminary data index table; Step S52: Optimize the weights and configure the retrieval rules for the preliminary data index table to obtain a searchable multi-source heterogeneous tsunami raw dataset.

5. The cloud computing-based tsunami disaster risk assessment method according to claim 1, characterized in that, Step S6 includes the following steps: Step S61: Perform model field filtering and structural reorganization on the searchable multi-source heterogeneous tsunami raw dataset to obtain the model mapping dataset; Step S62: Based on the model mapping dataset, perform unified encoding and structural association on the data semantic layer and logical layer to obtain a standardized data mapping table; Step S63: Perform model formatting and structured encapsulation on the standardized data mapping table to obtain a unified data model; Step S64: Perform cloud-based verification and service registration on the unified data model to obtain a reusable model service interface.

6. A cloud computing-based tsunami disaster risk assessment system, characterized in that, For executing the cloud-based tsunami disaster risk assessment method as described in claim 1, the cloud-based tsunami disaster risk assessment system comprises: The multi-source tsunami data access and preprocessing module is used to access and aggregate data from earthquake monitoring systems, satellite remote sensing systems, tide level monitoring devices, and historical tsunami event databases to obtain a multi-source heterogeneous tsunami raw dataset; the multi-source heterogeneous tsunami raw dataset is then processed for format standardization, time synchronization, and semantic annotation to obtain a unified data model; The unified data service construction and invocation module is used to establish and register data service interfaces based on a unified data model to obtain a data service set; and to invoke the data service set to obtain tsunami propagation simulation input data. The cloud-native parallel computing and task scheduling module is used to partition and containerize the input data of the tsunami propagation simulation to obtain a multi-node parallel computing task set; and to dynamically schedule and solve the multi-node parallel computing task set in parallel to obtain partitioned computing results. The partition result synchronization and data fusion module is used to synchronize and fuse the partition calculation results to obtain a global tsunami propagation result dataset. The risk indicator analysis and assessment module is used to perform statistical analysis and indicator extraction on the global tsunami propagation result dataset to obtain a set of tsunami disaster risk assessment indicators. The risk visualization rendering and interactive display module is used to render and dynamically display the tsunami disaster risk assessment index set, and obtain interactive risk distribution results.

7. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed, implements the cloud computing-based tsunami disaster risk assessment method as described in any one of claims 1-5.

Images