Overview
The Web-enabled Awareness Research Network (WARN) module provides subscribers with early notifications of tsunamis and earthquakes as part of the Smart Oceans BC initiative, utilizing some infrastructure already in place for Oceans 3.0 and incorporating some new instruments. Tsunami detection uses data from pressure sensors such as the existing Bottom Pressure Recorders (BPR) and some planned coastal radar devices. Earthquake detection uses new sensors providing p-wave data. While WARN is being implemented as a research/prototype with limited sensors and select early adopters for now, it is being designed to be expandable to greater numbers and types of instruments (such as coastal radar for tsunami detection) and subscribers.
WARN receives instrument data, and returns intrinsic event data (such as tsunami wave height, or earthquake magnitude and epicentre), which can be used in impact assessments by subscribers.
Key Features
Data acquisition
The data used by WARN is received from sensors, in real time, via the Data Acquisition module in Oceans 3.0. Pressure data for Tsunami evaluation originates from BPR sensors in the prototype. Data for earthquake assessment originates from accelerometers. Prior to analysis via WARN, the data is parsed and calibrated to pass data quality downstream for use in further analysis.
Single Sensor - Event Detection
Once data for a single sensor has been parsed and calibrated, it undergoes a varying number of data preparation and event detection algorithms, depending on the data type and the event to be detected. This stage evaluates sensor data and produces a detection event if various criteria are met.
Multiple Sensors - Correlation
Once an event has been detected on one sensor, then a correlation stage begins where events from multiple sensors are evaluated to produce an event confirmation. The sensors must be at at least three different locations to allow earthquake epicentre to be determined, and to give a level of confidence (earthquake and tsunami). Correlation produces an event confirmation.
Communication With Subscribers
Subscribers such as the Ministry of Transportation or utility companies will be able to subscribe to event notifications. The notification contains event confirmation intrinsic to an event, indicating the type of event, data relevant to event (eg. earthquake epicentre and tsunami magnitude). The subscriber would have some additional processing configured to their location and needs (ie. SHAKE and PREDICT), which would assess the potential impact and make the decision whether to send a warning.
Also see the following taken from the CANARIE Statement of Work:
Simulation/Replay/Testing
The WARN module provides replay function such that algorithms can be run against historical data to validate results. This is supported by storing a history of configuration values allowing the state of the WARN module to be re-created for any point in time. The WARN module will also support generation of test notifications to clients so they can test their impact assessment code.
Configurability
The WARN module will be configurable both scientifically and environmentally. It will allow its preparation or detection algorithms to be tweaked by ONC scientists at run-time. Any environment parameters will be configurable to allow the application to adapt to different geographical locations.
Tsunami Detection and Correlation
Pressure data from an instrument uses the following preparation algorithms before detection analysis:
- A detiding algorithm that removes the tidal component from pressure data allowing more accurate analysis by other algorithms.
It then is tested against three detection criteria in parallel:
- An algorithm that determines the ratio of short-term average to long-term average of pressure data. A ratio above a specified value indicates a possible tsunami.
- Kurtosis is used to evaluate if data exhibits non-Gaussian characteristics indicating a possible tsunami.
- A direct thresholding approach, checking that the residual signal after de-tiding does not exceed a threshold of concern.
A fourth algorithm detects Rayleigh waves (a type of earthquake wave) and prevents tsunami detection if in fact the pressure signal was caused directly by the earthquake. In the event that the earthquake subsequently produces a tsunami, the tsunami will of course be detected since it has a different pressure signature from a Rayleigh wave.
Tsunami Correlation within WARN will support recognition of tsunamis using inter-station travel time.
Earthquake Detection and Correlation
Earthquakes will be detected using accelerometers deployed around the province. Upon near-simultaneous detection at three or more sensors, results will be correlated, allowing location and magnitude for each earthquake to be calculated via triangulation.
Possible Future Enhancements
- Detection of tsunamis using coastal radar data
Unique Characteristics
Time sensitivity
The WARN module will be highly time sensitive as notification of geo-hazard events must be received by clients with enough time to prepare.
For earthquakes, the notifications must be sent no more than 2 seconds after detection. While subscribers very close to the epicentre may not receive notifications in time, this should provide adequate notification whenever possible. For tsunamis, the notifications will be sent within one minute after detection.