Presentation of David F. Merrick and Tom Duffy on the topic "Utilizing Community Volunteered Information to Enhance Disaster Situational Awareness" at ISCRAM2013
Utilizing Community Volunteered Information to Enhance Disaster Situational Awareness
1. The Florida State University
Tallahassee, Florida
Utilizing Community Volunteered
Information to Enhance Disaster
Situational Awareness
David Merrick
Tom Duffy
Center for Disaster Risk Policy, Florida State University
10th International ISCRAM Conference, May 2013
2. Traditional Inputs to Situational Awareness
There are a variety of traditional data inputs feeding
situational awareness, including:
Reconnaissance and damage assessments
Weather reports and forecasts
Geographical and population information
Partner organization reports
Private sector reports
This data is combined to create a conceptual
model of the disaster as it currently exists, as well
as projections for how it is going to evolve.
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
3. Community Volunteered Information
Why are response agencies interested in CVI and other
crowdsourced information?
Further, it provides a participatory channel for citizens
and survivors to help themselves and others.
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
4. Barriers to Adoption
There are significant barriers to adoption of CVI into situational
awareness processes and workflows.
Lack of authenticity and trust
Emergency and disaster managers swear by ‘ground truth’, and
they understand how difficult this can be to achieve. How can
an average Twitter user be trusted?
And yet… they want fast information.
Time intensive
Emergency and disaster managers already have limited
personnel resources to accomplish life safety and life
sustainment. Resources to process CVI is very limited.
Limited models or tools
Existing tools and models for working with CVI (primarily the
social media aspect) are aimed at commercial use.
The most powerful systems are too expensive for local EM use.
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
5. Overcoming Barriers
Lack of trust in CVI stems from a perception by
emergency managers of poor accuracy.
Increasing trust in CVI requires overcoming these
perceptions through research and case studies.
Further, the perception of poor accuracy stems from a
misconception of how CVI can be created and utilized.
FSU EMHS has established an ongoing research program
examining the validity of CVI.
Broad goals include:
Development of confidence indicators model for utilizing
CVI for situational awareness.
Development of best practices and workflows aimed at
practitioners.
Using CVI as an additional layer of information –
despite its imperfection.
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
6. Tropical Storm Debby (2012)
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
40 MPH Winds at Landfall
Maximum surge of 4.49 ft
at Cedar Key, Florida
Maximum rainfall of 28
inches in Sanborn,
Florida.
Minimal wind or surge
damage, but spawned
at least six tornados.
Major inland flooding
was the primary impact.
NWS Final Track - Debby
7. Tropical Storm Debby (2012)
Before landfall, EMHS began to capture Twitter data
based on keyword and hashtag searches.
“Debby”, “TSDebby”, “flwx”, “hurricane”, “TS Debby” and
others.
No geographical boundaries were utilized.
In a 7 day period, captured 36,317 tweets.
After analysis, it was determined that over 77% of the data
did not pertain to the tropical system.
Of the remaining 23% of the data, the concepts with the
largest frequency were “rain” and “flood/flooding”
Only 1.2% of the tweets were geocoded with latitude and
longitude coordinates.
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
8. Tropical Storm Debby (2012)
Additional real-time monitoring of Twitter traffic was
conducted during landfall and inland progression of the
storm.
EMHS used Hootsuite and keywords such as “TSDebby” and
“flooding” to identify data points.
Citizen reports were used in the State Emergency
Operations Center to get a very basic picture of which
counties were receiving flooding.
Significant analysis was not done on this data.
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
9. Hurricane Isaac (2012)
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
NWS Final Track
Isaac made landfall on
29 August 2012 with
maximum winds of 70
mph. It was classified
as a Category 1 storm.
Produced a surge of
over 14 feet in
Braithwaite, La.
Levees in Plaqumines
Parrish failed, causing
significant flooding.
Produced over 20 in, of
rain in parts of New
Orleans.
10. Hurricane Isaac Field Research
FSU EMHS deployed a team to New Orleans immediately
following Isaac’s landfall.
Their mission was to physically verify on CVI reports, assessing
accuracy of both community and government information.
This project served as a pilot for ongoing research.
Preliminary results include:
CVI that includes media (photos) was found to be more
accurate than CVI without media.
Citizens use terminology that is different than those who work
as a responder or in emergency management.
Information submitted may be accurate… but that doesn’t
make it relevant.
CVI is not comprehensive – and we cannot assume it will be.
It must be used as an additional information layer.
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
11. Isaac - Accuracy and Relevance
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
12. Isaac - Accuracy and Relevance
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
13. Isaac - Consistent Terminology
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
14. Isaac - Consistent Terminology
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
15. Ongoing Research
FSU EMHS will continue field research in the United States,
particularly as the 2013 Hurricane Season kicks off in
June.
We have refined our data collection system to allow us to
verify CVI more efficiently in the field.
Expand field research into observations of the State
Emergency Operations Center as well as County
Emergency Operations Centers.
Utilize the FSU VOST as a test platform for developing
models and best practices.
Integrate our developing models into the State of
Florida’s Division of Emergency Management through
ESF-14 (Public Information), ESF-5 (Planning), and ESF-18
(Business and Industry).
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany
16. Thank you!
Questions or Suggestions?
For more information:
David F. Merrick II
dmerrick@fsu.edu
Center for Disaster Risk Policy
Florida State University
644 Bellamy
Tallahassee, Florida 32306-2250
http://em.fsu.edu
850.644.9961
Merrick and Duffy, May 2013
10th International ISCRAM Conference,
Baden-Baden, Germany