1 Introduction
2 Related work
2.1 Data utilisation in multi-hazard early warning system
2.2 Semantic web technologies and high variety data management for multi-hazards
3 Landslip scenario
3.1 Scenario
3.2 Overall concepts
- Exposure—refers to people and environment which are living or located in landslide hazard prone area and are affected by landslide multi-hazard. In addition, environment can be classified to natural environment and built environment. The natural environment is all living and non-living things that occurred naturally (e.g. animals, river, forest, mountain, etc.). On the other hand, the built environment [6] is a combination of infrastructures and facilities produced by people as a core foundation in the community (e.g. house, school, road, bridge, electricity, water supply, etc.)
- Stakeholder—refers to people or organizations who have a stake in the landslide event. In the scenario, stakeholders are: (i) social media users who report landslide warning signs through their social media (e.g. Facebook, Twitter, Instragrams, etc.); (ii) data collectors and providers who deploy sensor devices in landslide hazard prone area and provide EO and urban data collecting from such devices to EWS for analysis. Data providers also include the third parties who collect data from sensor devices owned by the others. (iii) Decision makers who have responsible for conducting landslide hazard risk assessment using available social media data and EO and urban data. They make a decision based on result from Decision Support System and hazard risk management plan in order to inform people in risk area before the occurrence of landslide hazard.
- Event—refers to an occurrence which is related to a hazard. Additionally, hazard itself is also consider as an event. The hazard-related event is classified as pre-hazard event, post-hazard event and event during hazard. Since Early Warning System analyse EO and urban data to predict the potential of hazard in the area of interest, warning signs and anthropogenic processes are the majority of events in this scenario. In addition, a warning signs is an event that can indicate a possibility of hazards. An example of the warning sign is broken underground utilities which can be a warning sign for landslide hazard. An anthropogenic process refers to human activities which can induce hazards. An example of such activities is vegetation removal which induce landslide.
- Data Sources—refer to any sensors and data services that provide data to data consumers. These data sources have different capabilities to provide data. Sensor is a component that observes and measures physical phenomena and transform the observation and measurement into a human readable form. There are two types of sensor, physical sensor and human sensor. The data service is an application software that collects, stores and provides data from multiple devices. Several types of data sources are currently available to provide EO and urban data for multi-hazard applications.
- Decision Support Applications—refer to an integrated system that provide functionalities for stakeholders to monitor, forecast and predict, validate and assess hazardous events. In this scenario, EO and urban data collection system, data sources discovery services, hazardous event detection system and Early Warning System (EWS) are major components of Decision Support Applications. As a consequence, these applications enable stakeholders to take timely actions to reduce impacts of landslide hazard in advance. For example, once a landslide hazard is likely to be happened, a decision maker can make a decision based on information and knowledge from EWS to disseminate actionable warnings information to people in the landslide prone area.
4 Landslip Ontology
4.1 Landslip Common Ontology
- UrbanArea—defines concepts about urban area that prone to landslide including its basic elements. The urban area encompasses both natural resources (e.g. river, and mountain) and built environment, including infrastructure (e.g. road and railway), utility (e.g. electricity and tab water) and place (e.g. school, hospital, house and flat). Located in landslide prone area, these elements can be affected by landslide and other multi-hazards.
- NaturalHazard—defines a set of multi-hazards which can trigger land-slide hazard. This concept captures knowledge mainly on the interactions between landslide hazard and other multi-hazards (e.g. flood, earthquake, tsunami, and drought). In addition, the interactions between other multi-hazards are able to indicate landslide hazard.
- AnthropogenicProcess—defines a set of human activities that produce negative effects to landslide [8]. This concept also captures knowledge about the interaction with in the processes to provide direct and indirect indications of landslide hazards. In addition, the direct indications are the processes that are a trigger of landslide while the indirect indications are the processes that trigger other processes which trigger landslide. Moreover, the major indicators for anthropogenic processes are warning sign observed by a person.
- WarningSign—defines a set of incidents that can be an indications of landslide hazard, other multi-hazards and anthropogenic processes. The concept of warning sign is mainly focus on incidents which can be observed by a person. Such incidents are useful for landslide EWS in order to detect landslide precursors based on incidents reported in social network.
4.2 Landslip data sources ontology
- DataSource—is the main concept of Landslip Data Sources Ontology. A data source is any sensors or data services that provide observation data (e.g. physical sensor, human sensor and data service). DataSource defines a set of comprehensive information related to observation and data sources metadata which are the details of data sources.
- Observation—defines a set of observed properties (EO and urban data) which are used to observed features of interest related to landslide hazard. The examples of observed properties are soil moisture, soil movement, rain, earth quake magnitude, temperature, humidity, and wind speed. These observed properties are accessible to EWS via data sources.
- DataSourceMetadata—defines a set of information which are necessary for data acquisition process. This concepts is comprised of four groups of information profile: (i) observation profile—a set of observed properties provided by a data source; (ii) observed property profile—provides information about data type, feature of interest, and phenomenon time; (iii) sensor profile—provides information about type of sensor, feature of interest, and list of event to be observed; (iv) service profile—provides information which can be used to access a service (e.g. service type, end-point, provider); and (v) provider profile—provide the information about data provider (e.g. provider name, contact address).
4.3 Ontology metrics
Feature | Value |
---|---|
No of classes | 98 |
No of properties | 26 |
No of individuals | 30 |
No of axioms | 462 |
DL expressivity | ALCH(D) |
5 System architecture
- data sources layer—consist of a number of data sources provided by various data providers. Data sources collects EO and urban data from physical sensors deployed in landslide prone area. These sensors observe or measure properties of landslide and other earth observation which can be use to indicate landslide hazard. The data sources are accessible through a variety of methods (e.g. REST API, RDBMS, WSN) depending on data source providers. Moreover, data from social medias is also considered as data sources in this layer.
- data sources discovery layer—maintains the Landslip ontology which represents knowledge of landslide and data sources in a triplestore. It also provides data sources registry which store data sources metadata, including metadata for sensor, service and observation. Furthermore, there are a number of functionalities provided by this layer which allow users to (1) publish data sources; (2) search for potential data sources; and (3) indicate landslide hazard using warning signs. These functionalities are accomplished based on knowledge of landslide and data sources provided by the Landslip Ontology. In addition, the functionalities provided by this layer is accessible through data sources discovery service APIs which are available in form of RESTful Web Services API.
- hazard applications layer—provides client APIs to access the functionalities offered by the data sources discovery layers. In addition, the client APIs are design for both data provider and data consumer. Here, data provider can user the client API to register their data sources along with data sources metadata. On the other hand, data consumer uses the client API to search for potential data sources based on landslide warning sign.
6 Evaluation
Competency questions | |
---|---|
Q1 | What other hazards are likely to happen when hazard H has happened? |
Q2 | What is the probability of an event E occurring when warning sign W has been observed? |
Q3 | What is the probability of an event E occurring when a set of warning sign, W1, W2, W3, Wn have been observed? |
Q4 | Is warning sign W an indicator for landslide L? |
Q5 | What are observed properties that can be used to verify landslide when a warning sign W is observed? |
Q6 | Identify the data sources and their metadata required to observe a set of hazards (H1, H2, H3, Hn) |