Lessons from the Warning Project

Eve Gruntfest ecg@uccs.edu Kansas City January 22, 2009

The Big Thompson Flood in Colorado July 31, 1976

o 140 lives lost – 35 miles from Boulder

o Studied the behaviors that night o Who lived?o Who died?o Led to detection &

response systems

Focus on flash floods & warning systems

Our 2003-2008 National Science Foundation project aimed to

o Evaluate impacts of o Demographic

changeo New & different sources of

informationo Test

conventional wisdom abouto False alarms/

close calls

Evaluate previous trauma experience & warning perceptions

What we know about warnings – Public response components

o Hear/receiveo Understando Believeo Personalizeo Decide to acto Respond

Reinforcing what Havidan said this morning: The warning process is complex

Why Austin & DenverSimilarities

LargeGrowing cities

Diversepopulations

Vulnerable toflash flooding

Warning project methodology

o Survey Development o 1 yearo Input from officials & hazards researchers

o Survey formato Likert scale & true/falseo Demographic questionso Experience with flash floods & traumao Surveys in English & Spanish to selected

respondents

o Survey is available – for follow up studies

Mail survey

o 6000 surveys sent to residents in or near the floodplain

o 1017 surveys returned

o Where do people get their weather information?

o Best way for officials to warn you about a flash flood?

All sources of weather information used

81%

44%

11% 9% 10%

1%

68%

42%

14%

4%

36%

75%

90%

5%

25%

75%

92%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Televi

sion

Enviro

nmen

tal c

ues

Local r

adio

sta

tions

The W

eath

er Chan

nel

Inte

rnet

NOAA wea

ther r

adio

Wea

ther

Bug

Other

Cell P

hone

Denver

Austin

n=935

Most important source of weather information

46%

21%18%

6%2%

5%

49%

15%12% 12%

7% 4%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Local T

V

Enviro

nmen

tal C

ues

Local R

adio

The W

eath

er C

hannel

NOAA Wea

ther

Rad

io

Inte

rnet

DenverAustin

N=945

A flash flood warning indicates a more serious threat than a flash flood watch

0%10%20%30%40%50%60%70%80%90%

100%

Agree Disagree

92%

8%

n=1031

I take flash flood warnings seriously

0%10%20%30%40%50%60%70%80%90%

100%

Agree Disagree

n=1017

92%

8%

The best way for officials to warn you about a flash flood?

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Sirens Phone TV Cometo door

Radio NOAAradio

Email Cellphone

2:30 am11:00 am5:00 pm

N=1020

Is overwarning a problem: False alarm issues --”cry wolf” may not be a major problem

• People prefer more warnings even if there are more false alarms

• No measurement of “close calls” & “near misses”

• Events occur but not exactly in the warning areas or with exactly the predicted intensity

• How about new categories? New

metrics?

Officials are too sensitive to the possibility of flash flooding

N = 1031

86%

14%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Agree Disagree

Realizing it’s difficult to predict flash floods, I prefer more warnings even if there are more false alarms or close calls

N= 1047

78%

22%

0%10%20%

30%40%50%60%70%

80%90%

100%

Agree Disagree

Austin by GenderRealizing it’s difficult to predict flash floods, I prefer more warnings even if there are more false alarms or close calls

X² =4.150, p <.05

82%

18%

74%

26%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Agree Disagree

Male

Female

Extreme speed of

watershed responses

Extremely short lead-

time for warnings

New ways of representing what we know – include hydrology - Isabelle Ruin new time/space analysis

Challenging our assumptions - For flash floods – shorter lead time may be better than longer lead time

o Smaller area under warning

o Is there a best leadtime?

o Concerns with “too much leadtime?”

o Are these warranted?

We need new metrics –Socially relevant verification Forecast verification is difficult in rural areas

Prairie dogs don’t answer the phone

Since not all meteorological hazards are created equally

---What are acceptable levels of risk?

Infrastructure is aging! (levees for category 2,3,4,5?)

--- How do we measure warning success? If 20 people die in Greensburg, KS – warnings can still have saved hundreds

I would drive through an intersection with six inches of water running across it

4%

38% 39%

24%

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

Strongly agree Agree Disagree Stronglydisagree

63% say they would NOT DRIVE through it

Is this good or bad news?

Tailor message for local hazards --Maricopa County (Phoenix, AZ)

o Floodwaters can conceal a damaged roadway

o Flash Floods rarely last more than an hour

o Don’t trade an hour for a lifetime

Challenge of confronting ads from car companies

How to convince people they are better Wet than Dead?

Ad says: Warning: use the cup holders at your own risk

Behavioral survey project

Observe driver’s behavior at low water crossings in Texas

Quantitative survey• Use of video• car counting

Qualitative survey• Use Youtube video,

travels log & in-depth interviews

http://70.253.207.10/view/index.shtml

WHAT WERE THEY THINKING? USING TO OBSERVE

DRIVER BEHAVIOR CROSSING FLOODED ROADS

2009 Geography Master’s thesis by Cedar League

INTENTIONAL/SITUATIONAL

o Intentional drivers: purpose was to film the flood water, or to drive in the flood water (for fun). 59% (n=31)

o Situational drivers: purpose of trip was based upon their current situation, like driving to or returning from work. 41% (n=21)

Purpose of Trip

SOURCES OF WEATHER INFORMATION

Are warnings always possible? Do they always make THE difference?

Sirens – Technical capability for smaller than county warning – But whole county hears sirens?

West Nile Virus study epidemiology-

awareness example

Nobody identifies themselves as “elderly”

o Folks >50 years old most vulnerable to severe manifestations

o Campaign for elderly

How do we address nuisance events vs catastrophic events? Height of the 1997 Fort Collins flash flood

Warning project findings

•Weather information requirements of each user community are highly specialized

•The weather research community has not focused on the individual needs of specific user communities

•Existing social science studies barely scratch the surface – more comprehensive studies must be done to have more confidence in policy change based on findings

In 2018 – yet another decade from now…When we meet – we must seeEasy natural collaborations between physical scientists, planners, engineers, broadcasters, emergency managers, social scientists & others

Fewer deaths & reduced losses from weather events & a better understanding of how people use weather information

Las Vegas Billboards

Warning project publications Environmental Hazards 2007 -- Vol 7

o C. Benight, E. Gruntfest, M. Hayden, L. Barnes Trauma and short-fuse weather warning perceptions

o S. Drobot C. Benight, E. Gruntfest Risk factors for driving into flooded roads

o M. Hayden, S. Drobot, S. Radil, C. Benight, E. Gruntfest, L. Barnes Information sources for flash flood warnings in Denver, CO and Austin, TX

o I.Ruin, J-C. Gaillard, C. Lutoff How to get there? Assessing motorists’ flash flood risk perception on daily itineraries