gianluca correndo, simon crowle, juri papay and michael boniface | enhancing marine industry risk...

Enhancing Marine Industry Risk Management Through Semantic

Reconciliation of Underwater IoT Data Streams

14th September, 2016

Correndo G., Crowle S., Papay J., Boniface M.[gc,sgc,jp,mjb]@it-innovation.soton.ac.uk]

12th International Conference on Semantic Systems, Leipzig University

Professor Michael Boniface

© University of Southampton IT Innovation Centre 2016

Port of Leixões, Porto

© University of Southampton IT Innovation Centre 2016

Environmental impact Periodic maintenance

License compliance Accidents and extreme events

Risk managementMaintaining knowledge about the interplay between humans and ecological processes is an essential for risk management

© University of Southampton IT Innovation Centre 2016

DecisionMaker

Surveyor Analyst

InformationService Provider

MarineInformationEcosystem

Risk Management

Environment Data Acquisition Risk Analysis

Data Search, Aggregation, Fusion

Information Ecosystem

© University of Southampton IT Innovation Centre 2016

Hazard map for marine navigation channel based on sediment transport analysis

Hazard Maps

© University of Southampton IT Innovation Centre 2016

Underwater Internet of Things

© University of Southampton IT Innovation Centre 2016

Periodic maintenance

Modem

Sensors

Sonar

Camera

testbeds

Concept and Objectives

• Turn environmental and sensor data (including Underwater IoT streams) into knowledge – provide support for data discovery, integration, fusion,

presentation to communities and commercial teams

– use explicit semantics and discoverable linked data

– align with OGC Standards, tools, and Array DBs supports data fusion and visualisation

© University of Southampton IT Innovation Centre 2016

01/16 01/17time

Survey Data

Simulation Data

Hazard Maps

Fused Data

Today PlannedSurvey

PlannedSurvey

Maintaining situational awareness

© University of Southampton IT Innovation Centre 2016

• Portal to query, resample, aggregate and

process data o Import/Export binary files (e.g. GTIFF)o Metadata management and semantic

reconciliation (e.g. NASA SWEET)o Map layer dissemination (e.g. OGC WMS) o Fusion of experimental data from AUVs

with third party data sets

• Linked Data interface for access to data

on the webo Discoverability and semantic querying of

the metadata assetso Available in RDF/XML and Turtle

• Integration with SUNRISE GATE for UAV

data collection

EXPOSURES Capabilities

© University of Southampton IT Innovation Centre 2016

EXPOSURES Architecture

© University of Southampton IT Innovation Centre 2016

Semantic Reconciliation 1/2

• netCDF is a self describing data format for Earth observations (physical quantities)• provides a data model to describe what’s inside• sets of array data

• File’s are only a small piece of information in a wider context• same for every OGC WCS compliant server

• Need to reconstruct the link between the single dataset and the wider context• management of metadata is key in exploiting data assets

(retrieval, sharing, reconiliation)

© University of Southampton IT Innovation Centre 2016

NetCDF data model (EO data)

DataSetAttributeDimension

Variable

Time

Latitude

Longitude

http://mmisw.org/ont/cf/parameter/sea_surface_temperature

SST = f(time, latitude, longitude)

© University of Southampton IT Innovation Centre 2016

• NetCDF organises the data storage– small but well defined vocabulary to define metadata– function based interpretation of datasets

• Attributes, dimensions, and variables are strings– makes it difficult to reconcile such representation with external

data sets– limited support for federation of data providers– limited support for semantic interoperability

NetCDF data model

© University of Southampton IT Innovation Centre 2016

Semantic Reconciliation 2/2• We adopted the netCDF data model

• Metadata encoding into RDF and disseminated as Linked Data

• Alignment towards a comprehensive domain ontology• NASA SWEET (earth and environmental terminology)

• Exploitation of ontology alignments to support data discovery and retrieval

© University of Southampton IT Innovation Centre 2016

Linked Data Dissemination• Rationale

– existing standards (OGC WCS) allow data interoperability but do not promote discoverability or semantic interoperability

• Once semantically aligned the meta data can be made part of a bigger data ecosystem• i.e. Linked Data cloud

• Necessary for a broad range of marine and environment applications

© University of Southampton IT Innovation Centre 2016

Application Trial• Trial at Porto harbour explored a new data value

chain for the marine industry– routine evaluation of the current state of the seabed– surveyors with UAVs collect seabed bathymetry, water current velocities,

and spectral data for water turbidity

• Data ingested into EXPOSURES– create coherent, rasterized plane of environmental values (such as

seabed height or water temperature) – highly clustered data points sparsely distributed over AOI– interpolation and clipping functions applied to deal with missing and

extraneous data points

• Multiple surveys in areas of interest ingested and fused to produce for a more complete view

• Discovery and visualisation of aggregated data accessible via Linked Data

– visualizations of accretion/deletion of the seabed sediment profile– Geospatial temporal data queries

• Fusion functions prepare data sets for simulation– predict ecological processes, impacts and what if analysis– reduce data preparation time

© University of Southampton IT Innovation Centre 2016

Trial Results

© University of Southampton IT Innovation Centre 2016

Simulation Risk Map – Open TelemacObservation Risk Map – QGIS)

Summary

• EXPOSURES is a service platform supporting semantic alignment, geo-spatial fusion and linked-data access to underwater IoT data

• Cost and complexity of developing marine applications is reduced– harmonization of temporal and spatial resolution (resampling and interpolation), array based

data composition– data reconciliation of environmental properties– map layer visualization

• OGC and W3C linked data interfaces allow knowledge of marine environments to be curated, published and shared by communities and commercial teams

• Future work aims to extend the services to support a wider range of geo-spatial fusion capabilities

– automating semantic annotations of simulation input data within analytical workflows– exploring commercial business models for marine information services

© University of Southampton IT Innovation Centre 2016

Contacts & QuestionsMichael Boniface, Gianluca Correndo, Simon Crowle, Juri Papay

University of Southampton IT Innovation Centre [mjb, gc, sgc, jp]@it-innovation.soton.ac.uk

This research has been supported by the SUNRISE project under the FP7 framework, agreement number 611449.

© University of Southampton IT Innovation Centre 2016