MITIGATION OF THE IMPACT OF TORNADOES IN THE CANADIAN PRAIRIES

BY:

SAMANTHI W. DURAGE (PhD)

PROFESSOR S.C. WIRASINGHE (PhD, PEng)

PROFESSOR JANAKA RUWANPURA (PhD, PEng)

DEPARTMENT OF CIVIL ENGINEERING

SCHULICH SCHOOL OF ENGINEERING

UNIVERSITY OF CALGARY

CANADA

12th Annual Canadian Risk and Hazards Network SymposiumCoast Plaza Hotel, Calgary, Alberta

Disaster Mitigation in Canada

In Canada, approximately 80% of disasters are due to extreme weather

events such as tornadoes, hurricanes, hail storms etc (Hwacha, 2005).

Canada is gradually shifting from the ways governments have historically

approached disasters, through response and recovery methods, to

mitigation strategies (Emergency Management Act c.15, 2007).

Canada’s National Disaster Mitigation Strategy (PSC, 2010) highlights the

need to “apply and promote scientific and engineering best practices in

order to build a knowledge base for sustainable, cost-effective mitigation

decisions that contribute to community resiliency”.

2

Worldwide Occurrence of Tornadoes

3

Source:(Goliger and Milford,1998)

Tornadoes occur on both hemispheres between the latitude 20⁰ and 60⁰

but predominantly over the territory of the USA.

The overall yearly average number of tornadoes for the past decade is

1,274 in USA (NOAA,2011).

Source : Environment Canada

Tornadoes in Canada

According to Environment Canada, which is the authority responsible

for tornado detection and warning, an average of 43 tornadoes per

year occur across the prairies provinces and about 17 occur across

Ontario and Quebec.

Source: (Grosvenor et al, 1998)4

Pre-disaster Stage

Deaths and injuries could be minimized by taking appropriate actions at the pre-touch down phase of a tornado.

Mitigation of the Impact of Tornadoes

5Picture Courtesy: Public Safety Canada

Research Objective

To study, analyze, model, simulate and propose improvements to plans and systems to mitigate the impacts of tornadoes in the Canadian Prairies

Statistical analysis of historical Canadian Prairie data on tornadoes

Network modeling and simulation of the tornado detection, warning and

communication (TDWC)network

Stated preference analysis of how Calgary households and vehicle-drivers will take

protective measures (or not) in response to tornado warnings

Analysis of the total time consumption for warning, communication and initiation

of protective measures

Analysis of false warning and missed events in the Canadian Prairies

Recommendations for the stakeholders involved in the TDWC process6

7

1. STATISTICAL ANALYSIS OF THE TORNADO DATABASE

y = 0.3481x + 13.736R² = 0.2664

y = 13.822e0.0127x

R² = 0.2826

0

10

20

30

40

50

60

70

80

90

0 20 40 60 80 100

No

of

torn

ado

es

Year(1921-2011)

No of Tornadoes Reported

No of TornadoesReportedExpon. (No of TornadoesReported)Linear (No of TornadoesReported)

y = 0.0449x + 1.5087R² = 0.9632

0

1

2

3

4

5

6

7

1921 1931 1941 1951 1961 1971 1981 1991 2001 2011

Po

pu

lati

on

(In

Mill

ion

s)

Year (1921-2011)

Population vs. Time

Population

Linear (Population)

Model Development for the Tornado Time Trend

8

Regression Model+Time Series Model

Tornado Time Trend

This pattern shows a decreasing trend in the number of tornadoes

observed in the last two decades.

The downward trend may precede an upward trend in the number of

tornadoes.

It is noteworthy that, in the neighbouring US, a high variability can be

observed in tornado occurrences in the same period (Brooks et al.,

2014).

This regional pattern may be a climate change signal in how

tornadoes occur presently (Elsner et al., 2015).

9

2. EVALUATION OF THE WARNING COMMUNICATION AND RESPOSE SYSTEM

Objective:

To compare the total time consumption for warning, communication and initiation of protective measures with the warning lead time

Warning issuance to the warning receipt point

(Network Simulation)

+

Warning receipt point to the completion of protective measures

(Tornado Survey)

Curve fitting procedures required for this analysis used EasyFit software

10

Tornado Detection, Warning and Communication Network

11

Network Simulation – Monte Carlo Simulation

12

Stated Preference Survey

Nearly 500 Calgarians took part in the online survey and provided information on how they would respond to tornado warnings after receiving them.

The respondents were asked to assume that they received a tornado warning; and, their intended responses when at home and driving were collected separately.

13

Profile of Survey Respondents

Variable Response Categories Percentage

Gender Male 66.0Female 34.0

Age Below 30 13.5Between 30 and 50 52.8Above 50 33.7

Dwelling Type Single Detached Dwelling 74.6Other 25.4

Household Size One 12.8Two 32.3Three or More 54.9

Presence of School Aged Children

Yes 33.5No 66.5

Presence of People with Reduced Mobility

Yes 6.2No 91.7Not Answered 2.1

Household Income Less than $50,000 9.3$50,000 - $120,000 36.8Above $120,000 40.1Not Answered 13.8

Level of Education Up to High School 7.1Training after High School 32.8Undergraduate Degree 38.5Postgraduate Degree 19.0Not Answered 2.6

Household Responses

Warning Sources Protective Actions

Household LevelRanking Average

(1-6)

Go to a safer area in the basement 1.47

Lie down in a bathtub 3.20

Go to a safer building within the neighbourhood

3.39

Drive away to avoid the threat 3.70

No action 3.79

Other 5.46

14

0 1 2 3 4 5

Local radio

Emergency weather radio

Social media

From neighbours

Television

Mobile text alerts

Weather websites

Visible environmental cues

Call from a trusted person

Please rate each likely source of warning to your household in case of a tornado

(1-Least likely , 5-Most likely).

Overall Time Consumption

15

Warning issuance to the warning receipt point

(From the network simulation)

Warning receipt point to the completion of the protective action

(From the tornado survey)

Overall Time Consumption

There is around 25% chance that the protective actions can be completed by a household within 10 minutes from the warning issuance point.

Increasing the time by 5 minutes gives more than 70% chance for a household to complete their protective actions.

16

3. PRAIRIE DATA ANALYSIS

A true warning for a tornado is a clear communication to the public to move to safer places prior to an actual occurrence.

A false warning can be considered to be a situation when the public is warned about a tornado and one actually does not occur.

A missed event is a situation where a tornado touchdown occurred without an advance warning being issued.

This status quo infers the correct detection of the situation that there is no tornado potential within a thunderstorm, thus, no warning is required.

Tornadoes Observed

Yes (T) No (T)

Tornadoes Forecasted

(or Warned)

Yes (W) p True Warning r False Warning

No (W) q Missed Event s Status Quo

17

Analysis of Tornado Warnings in the Canadian Prairies

18

Probability of

True Warning

P(T/W)

Probability of

False Warning

P(T/W)

Probability of

Missed Event

P(T/W)

Probability of

Detection

P(W/T)

Probability of

False Detection

P(W/T)

12.6% 87.4% 10.7% 39.8% 35.4%

True Warning, False warning, Detection Probabilities given a Severe Weather Bulletin

Venn Diagram of tornado warning, occurrence records from 2003 to 2012

4. HOUSEHOLD DECISION TREE FOR A TORNADO

19

Do not

Respond

Respond

Tornado

No Tornado

Tornado

No Tornado1-p

p

1-p

p

Warning

U2 =Property damage +

Some impact on people

Disutility

U3=Property damage +

Major impact on people

EventDecision

p =Probability of a tornado given that a warning

has been issued

U0=0

U1=Inconvenience+ Injuries

during response phase

The expected disutility of responding to a warning E (R) = p(U2) + (1-p)(U1)

The expected disutility of not responding to a warning E (R′) = p(U3)

The household will choose to respond if E (R′) >E(R)

Household Decision Tree for a Tornado

20

Inequality

(U3-U2) - the additional disutility or the consequences of not responding to a tornado warning

U1 - Negative consequences of responding to a false warning

-When (U3-U2) is higher even a low probability of a true warning p is sufficient to trigger taking protective actions. -When U1 is higher, it is necessary to have a higher value for p to initiate response actions

Case (i) (U3-U2)> U1 ; >1 RHS of inequality < 0.5

Case (ii) (U3-U2)< U1; <1 RHS of inequality > 0.5

(Since U1 is small and U3-U2 is high, it is unlikely that case (ii) will occur.)

5. WARNING DECISION TREE FOR A TORNADO

U3 =Property damage + Major impact on people

Do not

Respond

Respond

Tornado

No Tornado

Tornado

No Tornado1-p2

p2

1-p2

1-r

r

p2

Warning

No Warning

Tornado

No Tornado1-p1

p1

U2 =Property damage + Some impact on people

Disutility

U3=Property damage + Major impact on people

Public Response

Decision

r = Probability of respondingp1 =Probability of a tornado given that no warning has been issued P(T/W’)p2 =Probability of a tornado given that a warning has been given P(T/W)

U0=0

U0=0

U1=Inconvenience+ Injuries during response phase

21

The expected disutility of giving a tornado warning E (W) = rp2(U2) + r(1-p2)U1 +(1-r )p2 (U3)

The expected disutility of not giving a tornado warning is E(W)= P(T/W)(U3) =p1(U3)

The forecaster should choose to issue a warning if E(W)> E (W)

Fundamental inequality of decision making for tornado warningsr > { 1 - [ P(T/W’) / P(T/W) ] } / [ 1 - (U2/U3) ]

Case (i) P(T/W)>P(T/W) or (Missed event probability > True warning probability) ; RHS is always negative since U2<U3

No matter what response probability is expected from the public, the forecaster should choose to issue a warning.

Case (ii) P(T/W)< P(T/W) or (Missed event probability < True warning probability); RHS is positiveThe response probability has to be higher than a certain positive value to justify a warning.

22

Warning Decision Tree for a Tornado

6. RECOMMENDATIONS

23

Partner Recommendations

SPC -Check the sufficiency and efficiency of the technological and human capacity to detect

tornadoes and take remedial measures

-Implementation of a group of spotters to get ground-truth information

-Develop interactions with the local emergency managers

Promote the use of Weatheradio application as the primary warning source

CEMA -Conduct annual information sessions and drills to improve the awareness and

preparedness at the individual level, institution level and the community level

-Develop interactions with the SPC, spotters and the public to get tornado information

and activate the AEA

-Promote various communication media including the Internet, social media and

Smartphone applications that can reach a diverse population with different preferences

AEMA -Strengthen the AEA system to verify tornado information at the local level

RECOMMENDATIONS (cont’d)

24

Partner Recommendations

Schools -Practice tornado drills in the Springs season

-Improve the awareness of parents regarding school protective measures

ROC -Initiate the use of VMS for severe weather warnings including tornadoes

-Educate drivers on how to respond to a tornado emergency

-Study of traffic management technologies to assist in responding to a tornado

warning

Media -Educate the public by facilitating discussions about tornado preparedness and

response in the Spring and Summer seasons

Police -Be ready to respond once a tornado touchdown is reported

Public -Improve awareness about environmental cues of tornadoes, weather alerts,

warnings and protective actions

-Develop a family preparedness plan in responding to a tornado

Publications

Durage, S.W., Wirasinghe, S.C., Ruwanpura, J.Y. (2011). Comparison of the Canadian and U.S. Tornado Detection and Warning Systems, Natural Hazards, Vol. 66, No. 1, pp. 117-137. [Published]

Durage, S.W., Kattan, L., Wirasinghe, S.C., and Ruwanpura, J.Y. (2014). Evacuation Behaviour of Households and Drivers during a Tornado, Natural Hazards, Vol. 71, No. 3, pp. 1495-1517. [Published]

Durage, S.W., Wirasinghe, S.C., and Ruwanpura, J.Y. (2015). Decision analysis for tornado warning and evacuation. Natural Hazards Review. DOI 10.1061/(ASCE)NH.1527-6996.0000195 . [Published ]

Durage, S.W., Wiraisnghe, S.C., Ruwanpura, J.Y., Kattan, L and Mashall,S. (2015). Canadian Prairie Tornadoes - Preplanning for Warning Issuance & Initiation of Protective Measures, International Journal of Disaster Risk Reduction. [Accepted for publication]

Durage, S.W., Wirasinghe, S.C., Ruwanpura, J.Y. (2014). Tornado Mitigation Network Analysis and Simulation. International Journal of Disaster Resilience in the Built Environment. Manuscript ID IJDRBE-10-2014-0072 [Under Review].

25

Acknowledgements

Calgary Emergency Management Agency (CEMA)

Environment Canada - Prairie and Arctic storm prediction Centre

Natural Sciences and Engineering Research Council of Canada (NSERC)

Urban Alliance

Storm Prediction Center –USA

Calgarians

26