Presentation on theme: "Preliminary Look at Public Perceptions of and Responses to Warnings in the 27 April 2011Tornado Outbreak Kim Klockow University of Oklahoma Elise Schultz."— Presentation transcript:
1Preliminary Look at Public Perceptions of and Responses to Warnings in the 27 April 2011Tornado OutbreakKim KlockowUniversity of OklahomaElise Schultz and Stephanie Mullins UAHuntsville18 October 2011National Weather AssociationBirmingham, AL1
3Broad-brush overview of the event from a warnings/storm tracks perspective. Some places had upwards of 12 warnings (just at their location, not even considering county). Different practices at two WFOs, different average lead-times. Final analysis will try to identify any observable effect (may be really challenging).Brief description of where I went, sampling method, data collection method. Note the flexibility to formulate new hypotheses, necessary given the levels of fatality observed. 71 interviews (mainly members of the public, 5 officials & broadcasters). ~ 30 seconds.Situation well-forecast. Lots of information out days before. Still a severely confusing situation to understand w/am convection, so let’s try to break down how people figured out what was going on, and what responses looked like.
4Public perceptions & response Note: Preliminary findings!Public was largely cognizant of anticipated severityCritically important: Built-up awareness, continued attentionRelation to April 3, 1974Optimism bias always presentWhat realistic response looks likeUncertainty reduction/calibration ideasThe last few minutesMix of information sourcesTV, radio* & NWR, sirens**, internet (NWS, private vendors)Family/friends, police scanners (local)The environment!Locally-situated understandings of hazards/science, or “folk science”A significant issue: local, situated understandings of hazards, or “folk science”On anticipated severity: Why built-up awareness was so important. Optimism bias = reduces tendency to take preventative care. Nobody ever thinks it will happen to them. This is approximately true (bias is somewhat small)…it’s extremely unlikely. Well-calibrated to the risk overall. What can we hope for, then? Awareness. Suspicion carried through an event that something just might happen to them. Odds increasing. People are thinking about this probabilistically, think in terms of odds.People know the low probability is low. Warning != “seek shelter immediately,” though most know it’s what they’ve been *taught*. They *learn* what reality is over many trials. Result? High certainty required for action. To me: comes from spatial awareness + personalization + belief. When does this come together? The last few minutes. But note… response is not only diving for shelter! It’s also preparing to shelter, which happens at much lower stages of certainty + high awareness.Radio = used more heavily as tornadoes approached (power outages, driving). Very important note about when radio audiences are listening critically, what they’re listening for.Maybe we shouldn’t fight the use of environmental signals – Try to tell people what they’ll see! Not just from a chopper, but on the ground. Chasers can be very useful!
5Spatial AwarenessWhat information, specifically, was sought? What is that *best* information to provide?Loops, trajectory cones, TOAs* (primary)Local information (primary)Each person had an action/response threshold.People don’t comprehend large tornado scaleFootage of tornadoes (secondary)Three sources of uncertaintyTrajectory of storm (where)Tornado yes/no (existence)Severity/strength (how bad)Key Idea: There are a lot of tradeoffs involved!Level of detailProvide clear, simple, and decision-relevant informationTechnical capabilities to provide detail without meaningful errors.So what’s most important? Helping them to gauge level of uncertainty via trajectory inference, timing estimation. Something they can monitor and update, like how we validate model performance.Info: Think about this from a cognitive perspective. What task are people performing? Uncertainty reduction IF you have their attention. You don’t always, in most cases, you have scattered attention/cognitive load. A “thin-slicing” mode for those familiar with the writing of Malcolm Gladwell, has specific properties. Among them the incredible need for noise reduction, ability to identify *most relevant information* declines. You must be careful with what you’re giving, especially in complex situations like this. A simple and most highly relevant set of information is much much better than thoroughly detailed, technical information! Difficult task: be specific (local), but not dramatically over-detailed.What the world looks like, approximately, to someone responding to a tornado threat.
6Suggestions from interviewees Fix county-based sirens so they’re a better cueTheir definition of a “false alarm” != oursTell people to put their shoes on or grab them before seeking shelterMany injuries sustained when leaving their safe place, rescues precludedTell ambulances to carry fix-a-flatBroadcasters: Be aware of radio audience!Comprehensible spatial references, scale descriptionsEspecially important when audience grows in the final minutes before a tornado hits, and nearing storm knocks out power: continued understanding is criticalDon’t linger too long on tornado video or a zoomed-in still imageKeep a map up so people know where the tornado is and is goingGive people a small inset map so they know where you’re zoomed in toShow loops so people can infer trajectory
7Cullman County Population: 80,500 Sirens only turned on for tornado warningsIncrease in reports of SLC types after 27thPublic on edgeTalks to public to advocate preparednessFacebook page starting Jan: ~3,000 “likes”Post other items (road status, etc) in addition to wx infoReduction in phone callsHams, fax, media, …
8Marion County Population: ~30,000 Public aware prior to 27th led to preparedness actions?Emergency response overwhelmedCommunication issues, office twitter account (~60 followers)Power loss – “was like we went back in time”Mass notification system: “wish list” ($$)22 community shelters in countyRecovery/response difficult w/ limited resources, diversified industryTuscaloosa vs. Hackleburg (pop. 1300)
9Lawrence County Population: ~35,000 Sirens coverage limited 3 main populated areasMoulton, Town Creek, and Browns FerryTalking siren (Moulton) used on April 27Once power lost = no sirensNWS EMA and in warning text – good job of emphasis multi rounds, but public let guard downOverwhelming new interest in sheltersFEMA assistance program5 community shelters approved, 8 more applied forOnly one now is in Moulton (pop. ~3200)Facebook account: started after 27th, 1500+
10Mt Hope (Lawrence Co) Residents Left: Shelter residents, their friends & neighbors survived inBottom left: entrance way (note pole that blocked way out)Below: interviewing residents/friends who survived, progress on new house with shelter in background
11Madison County Population: ~320,000 Close connection to NWS Most all EMs have thisLocation helps Madison hereSirens used only for ][ warningsCan only alert entire countyProblem of a perceived higher FARWebsite, media, Facebook pageAs with all EMs, strong connection with HAMsDifficulties with allocation of resourcesSharing amongst other counties good, but a challenge when multiple rounds of svr wx impact the region over the entire day
12DeKalb County Population: ~70,000 Good model for others Sirens sectioned offHelps to limit an increase in perceived FARMass notification systemPublic signs up (free)Phone calls, text messagesFacebook page: 2000+NWS forecasts, warnings, statementsRoad, power, other conditions2 way flow of info, reduction in phone callsOn 27th though, power loss…Post 27th: SLC reports
13Rainsville (DeKalb Co) Resident This was home, and yet they are grateful, hopeful, and optimistic.Above, left: aftermath to home, garage, vehiclesRight: shelter where family survived
14Now vs. Need: Localized warnings 27th well forecasted SBWs good, but sirens not tied to them in most areasIncreases “perceived” FAR27th well forecastedShorter term info to publicTV, Smartphones, NWR, etcSocial media presence – now in nearly all EMA officesAwareness high now, but for how long?Geospatial reference improvements