Transcript A WATERS Network Testbed

An Environmental Information
System for Hypoxia in Corpus
Christi Bay:
A WATERS Network Testbed
Paul Montagna, Texas A&M University Corpus Christi
Barbara Minsker, University of Illinois Urbana-Champaign
David Maidment and Ben Hodges, University of Texas Austin
Jim Bonner, Texas A&M University College Station
Acknowledgements



Funding for the CCBay Testbed comes from NSF.
Funding for data collection comes from Coastal Bend
Bay and Estuary Program, Texas General Land
Office, and the Texas Water Development Board.
Project teams are thanked for their contributions to
the emerging EIS system.



The Consortium of Universities for the Advancement of
Hydrologic Sciences, Inc (CUAHSI),Hydrographic
Information Systems (HIS) Project.
National Center for Supercomputing Applications (NCSA),
Environmental CyberInfrastructure Demonstrator (ECID)
Project.
WATERS Network.
WATERS Testbeds
Corpus Christi Bay, Texas
Testbeds in WATERS Network

WATer and Environmental Research Systems
(WATERS) Network:

A proposed networked infrastructure of
environmental field facilities working to promote
multidisciplinary research and education on
complex, large-scale environmental systems.




A network of instrumented field facilities
A facility that assists with and provides training on sensor
deployments, measurement campaigns, and sensor
development
Multidisciplinary synthesis of research and education to
exploit instrumented sites and networked information
An environmental cyberinfrastructure
Cyberinfrastructure (CI)







Computers
Networks
Archives
Grid services
Collaboration services
Information technology services
Data management, mining, and visualization
services
Why Corpus Christi Bay (CCB)?

A good question:




Can we forecast hypoxia?
Existing long-term data sets
Existing sensor networks
Manageable place to prototype CI
CCBay Goal and Questions:

To observe, model, and understand hypoxia in Corpus Christi Bay
with advanced sensing and environmental information systems

Understand Hypoxia:



Integrate the Observing System:


Can data from different sensors be combined to depict hypoxic conditions in
real-time and guide sampling strategies?
Model the System:


How is hypoxia interrelated with dissolved oxygen dynamics,
hydrodynamics, and salinity?
How do engineered systems impact hypoxia?
Can hydrodynamic and salinity conditions occurring during hypoxic events
be successfully simulated using known mechanisms and/or or machine
learning (i.e., data mining)?
Build Environmental Information System (EIS):



How can the EIS for in Corpus Christi Bay be applied as a template for the
investigation of hypoxia at other locations?
Can cyberinfrastructure elements of a digital bay be adapted for other water
environments?
What data models best integrate observed and simulated information in
three-dimensional water bodies?
Sensors in Corpus Christi Bay
Sensors in Corpus Christi Bay
NCDC station
TCEQ stations
TCOON stations
Hypoxic Regions
Montagna stations
USGS gages
SERF stations
National Datasets (National HIS)
USGS
NCDC
Regional Datasets (Workgroup HIS)
TCOON
Dr. Paul Montagna
TCEQ
SERF
CC Bay Researchers Currently Cannot
Adapt Monitoring to Hypoxia Events


Oxygen data from continuous sondes are only
downloaded weekly
Other sensor data are available in near-real-time, but
correlations with oxygen levels have not been
quantified



For example, wind speed & direction, water surface level,
salinity, and temperature
Manual sampling should be increased when
probability of hypoxia is high, but researchers cannot
integrate diverse data and models to predict when to
mobilize
Cyberinfrastructure can create an information system
to enable near-real-time, adaptive monitoring
Solution is to Create a
Digital Watershed
A Digital Watershed integrates observed and modeled data from
various sources into a single description of the environment
Environmental Information
System Servers
Observations
Server*
GIS Data
Server
Digital
Watershed
Weather Server
Remote Sensing
Server
*Using the Observations Data Model (ODM)
Observations Data Model
ODM = Observations Catalog + Values Table + Metadata Tables
EIS Server Architecture

Map front end –


Relational database –


ArcGIS Server 9.2 (being
programmed by ESRI Water
Resources)
SQL/Server 2005 or Express
Web services library –

VB.Net programs accessed as a
Web Service Description
Language (WSDL)
Environmental CI Architecture
Integrated CI
Knowledge
Services
Create
Hypothesis
Data
Services
Workflows
& Model
Services
Obtain
Data
Analyze
Data &/or
Assimilate
into
Model(s)
Supporting Technology
MetaWorkflows
Link &/or
Run
Analyses
&/or
Model(s)
Collaboration
Services
Discuss
Results
Digital
Library
Publish
Research
Process
CC Bay Near-Real-Time
Hypoxia Prediction Process
Anomaly
Detection
Replace or
Remove Errors
Update Boundary
Condition Models
Sensor net
C++ code
Data
D2K
workflows
Fortran
numerical
models
IM2Learn
workflows
Archive
Hypoxia Machine
Learning Models
Hypoxia Model
Integrator
Visualize
Hypoxia Risk
Hydrodynamic
Model
Water Quality
Model
Visualize
Hydrodynamics
Workflow Using
Cyberintergrator Development

Studying complex environmental
systems like Corpus Christi Bay
requires:



Coupling analyses and models
Real-time, automated updating of
analyses and modeling with diverse tools
CyberIntegrator is a prototype
technology to support modeling and
analysis of complex systems
CyberIntegrator
Event-Driven Architecture

What is an event?


When something noteworthy happens in
one component of the CI that should be
broadcast to other components of the CI.
Applications in the cyberinfrastracture
can produce or consume events.

For example, sensor anomaly detected,
or predicted hypoxia requires focused
manual sampling.
Sensor Anomalies


Sensors are not always reliable (see above wind
data), and real-time data can be difficult to check
by hand
We have developed machine learning anomaly
detectors

Being implemented with data services in
CyberIntegrator to automatically detect anomalies &
alert data managers
Event Architecture
Producer
Event Broker
Anomaly Detection
Event: Anomaly Detected
(JMS Broker)
Handle messages
and their distribution
Consumer
CyberIntegrator
Visualize anomaly
and previous ten values
Consumer
Event: Anomaly Detected
Detect anomaly in
data from Sensors
Consumer
System Tray
Notification App
Portlet
Notify user of anomaly
Visualize published
events
How Will All This Help
Researchers in CC Bay?

Consider the following scenario that
defines what could be enabled …
Hypoxia Alert



John Doe gets a page saying that hypoxic conditions
are predicted with 80% certainty in 24 hours
John logs into the CyberCollaboratory, where he joins
an ongoing chat with researchers (both local and
across the country), who also received the alert, and
are looking at the data and model predictions
The researchers agree that the predictions appear to
be reasonable given the current conditions


John mobilizes his research team to deploy detailed manual
sampling of the affected region the next morning
He uses the CyberCollaboratory to notify students & volunteers
from the local region who have indicated an interest in helping
with field sampling
Hypoxia Alert

When the samplers and crews are mobilized, the
data they collect are transmitted back to the data
storehouse



Model predictions made by CyberIntegrator meta-workflows
are updated automatically
Additional data needs are identified with CyberIntegrator
meta-workflows and are transmitted back to the crews
through event subscriptions
Others monitor visualizations of hypoxia in real time
and discuss implications in the CyberCollaboratory

Useful to:



Regulators & stakeholders
Researchers and students across the country
Interested public (fisherman, teachers, journalists)
New Paradigm

Cyberinfrastructure can enable nearreal-time adaptive monitoring,
modeling, and management of largescale environmental systems through:


Web services architecture to deliver
diverse data quickly and easily
Event-based cyberenvironments enable
users to easily link and adapt complex
models and analyses