The Perception of Volcanic Risk in Kona Communities from Mauna Loa and HualaLai Volcanoes, Hawaiki C.E

Available online at www.sciencedirect.com

Journal of Volcanology and Geothermal Research 130 (2004) 179^196

www.elsevier.com/locate/jvolgeores

The perception of volcanic risk in Kona communities from Mauna Loa and Hualalai volcanoes, Hawaiki C.E. Gregg a;Ã, B.F. Houghton a, D.M. Johnston b, D. Paton c, D.A. Swanson d

a Department of Geology and Geophysics, University of Hawaiki, 1680 East-West Road, Honolulu, HI 96822, USA b Institute of Geological and Nuclear Sciences, 69 Grace¢eld Road, Lower Hutt, New Zealand c University of Tasmania, School of Psychology, Bag 1-342, Launceston, Tasmania, Australia d U.S. Geological Survey, Hawaiian Volcano Observatory, P.O. Box 51, Hawaiki National Park, HI 96718, USA Received 9 April 2003; accepted 21 July 2003

Abstract

Volcanic hazards in Kona (i.e. the western side of the island of Hawaiki) stem primarily from Mauna Loa and Hualalai volcanoes. The former has erupted 39 times since 1832. Lava flows were emplaced in Kona during seven of these eruptions and last impacted Kona in 1950. Hualalai last erupted in ca. 1800. Society’s proximity to potential eruptive sources and the potential for relatively fast-moving lava flows, coupled with relatively long time intervals since the last eruptions in Kona, are the underlying stimuli for this study of risk perception. Target populations were high-school students and adults (n = 462). Using these data, we discuss threat knowledge as an influence on risk perception, and perception as a driving mechanism for preparedness. Threat knowledge and perception of risk were found to be low to moderate. On average, fewer than two-thirds of the residents were aware of the most recent eruptions that impacted Kona, and a minority felt that Mauna Loa and Hualalai could ever erupt again. Furthermore, only about one-third were aware that lava flows could reach the coast in Kona in less than 3 h. Lava flows and ash fall were perceived to be among the least likely hazards to affect the respondent’s community within the next 10 years, whereas vog (volcanic smog) was ranked the most likely. Less than 18% identified volcanic hazards as amongst the most likely hazards to affect them at home, school, or work. Not surprisingly, individual preparedness measures were found on average to be limited to simple tasks of value in frequently occurring domestic emergencies, whereas measures specific to infrequent hazard events such as volcanic eruptions were seldom adopted. Furthermore, our data show that respondents exhibit an ‘unrealistic optimism bias’ and infer that responsibility for community preparedness for future eruptions primarily rests with officials. We infer that these respondents may be less likely to attend to hazard information, react to warnings as directed, and undertake preparedness measures than other populations who perceive responsibility to lie with themselves. There are significant differences in hazard awareness and risk perception between students and adults, between subpopulations representing local areas, and between varying ethnicities. We conclude that long time intervals since damaging lava flows have occurred in Kona have contributed to lower levels of awareness and risk perceptions of the threat from lava flows, and that the on-going eruption at K|lauea has facilitated greater awareness and perception of risk of vog but not of other volcanic hazards. Low levels of preparedness may be explained by low perceptions of threat and risk and perhaps by the lack of a clear motivation or incentive to seek new modes of adjustment.

* Corresponding author. E-mail address: [email protected] (C.E. Gregg).

0377-0273 / 03 / $ ^ see front matter ß 2003 Published by Elsevier B.V. doi:10.1016/S0377-0273(03)00288-9

VOLGEO 2711 6-1-04 180 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196

ß 2003 Published by Elsevier B.V.

Keywords: Mauna Loa; Hualalai; Kona; lava £ow; vog; risk perception

1. Introduction paredness (Mileti and Sorensen, 1990; Perry and Lindell, 1990; Lindell and Perry, 1992; Mileti and Hawaiki’s coastal communities are subject to a Fitzpatrick, 1992; Mileti and O’Brien, 1993; Lin- wide range of frequent and infrequent natural dell, 1994; Nathe et al., 1999), but this link is hazards (Fletcher et al., 2002) ^ £oods, tsunami, often uncertain and tenuous (Lindell and Whit- storms, extreme weather, seismicity and volcanic ney, 2000; Johnston et al., 1999). The dissemina- eruptions (Hamnett et al., 1993, 1996). As a con- tion of information related to natural hazards sequence of rising population densities, increas- may promote hazard awareness among individu- ingly sophisticated facilities, and increasingly als in a community, but the widely maintained complex social and economic infrastructure (Bur- assumption that increasing an individual’s aware- by, 1998; Johnston et al., 1999), Hawaiian society ness of hazards will result in an increase in their is growing increasingly vulnerable to experiencing levels of preparedness for natural hazards is not detrimental consequences from natural hazard ac- always justi¢ed and has fallen well below expect- tivity. This vulnerability, coupled with the unpre- ations (Lindell and Whitney, 2000; Paton et al., dictability of natural hazards, in regard to nature, 2000). No direct relationship between awareness scale, timing, duration, and location, increases the and risk perception has been unequivocally dem- importance of increasing community resilience. A onstrated. To make matters worse, people con- key element of the latter concerns ensuring com- fronted with complex issues about which they munities and their members are well prepared. have a poor understanding, such as volcanic haz- One barrier to hazard readiness is incomplete ards, may transfer all responsibility for their pro- understanding of the factors that in£uence indi- tection to the requisite experts, e.g. the U.S. Geo- vidual and societal preparedness, and of how they logical Survey Hawaiian Volcano Observatory or react to outreach activities prior to these crises. Civil Defense agencies in the case of Hawaiki(Pa- Hazard mitigation strategies typically focus on ton et al., 2001a), with reduction in preparedness the physical attributes (e.g. the magnitude, fre- (Mulilis and Duval, 1995; Lindell and Whitney, quency and physical processes of natural hazards, 2000). These observations highlight the need to and on engineering and building design), but articulate people’s understanding of scienti¢c in- rarely consider the meaning this has for people formation and technical mitigation measures be- or its relationship to risk reduction behavior. fore embarking on a strategy that involves using The importance of this relationship is heightened this information to facilitate preparedness. Adopt- by the fact that technological mitigation measures ing this approach gives scienti¢c and emergency that increase objective safety may lead to over- management agencies the opportunity to correct con¢dence and risky behavior (risk homeostasis, misconceptions ¢rst. They can then develop risk Adams, 1995). Similarly, the dissemination of sci- communication strategies that build on communi- enti¢c information by experts within outreach ty knowledge, beliefs, needs and expectations programs that focus only on supplying accurate (rather than providing information that re£ects information to the public, without considering only the knowledge and expectations of the scien- how recipients will react to that information and ti¢c community). whether they will use it e¡ectively in the future, Several studies have measured understanding ironically can also reduce the perceived need for and perception of volcanic hazards and risk dur- preparedness (Paton et al., 2000). ing periods of quiescence (D’Ercole et al., 1995; People’s acceptance of their vulnerability can Johnston and Houghton, 1995; Perry, 1990), after increase warning e¡ectiveness and improve pre- a volcanic crisis (Murton and Shimabukuro,

VOLGEO 2711 6-1-04 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 181

1974; Kartez, 1982; Saarinen and Sell, 1985; impact than other hazards, because of the rela- Yoshii, 1992), and before and after a volcanic tively long durations of precursory and eruptive crisis (Johnston et al., 1999). However, few studies activity. For example, storms arrive with high have been conducted in Hawaiki, whose varied wind and rain but generally pass within a few ethnicity, wide cultural framework, and volcanism days, and tsunami pass after several hours have are unique in the world. elapsed. On the other hand, eruptions at Mauna While work on perceptions of risk or prepared- Loa have had durations of from less than 1 day to ness for volcanic hazards in Kona is absent from 4.5 years of almost continual activity during the the literature, some work has been done on summit eruption of 1872^1877, and £ank erup- K|lauea volcano (Fig. 1a) in the eastern portion tions have had durations from about 5 days to of the island (e.g. Lachman and Bonk, 1960; 450 days (Lockwood and Lipman, 1987). The Murton and Shimabukuro, 1974). This study most recent lava £ows from Hualalai may have evaluates the perceptions of volcanic hazards had durations of hours to months (Kauahikaua et from Mauna Loa and Hualalai volcanoes in al., 2002). Hazards with unpredictable, but poten- Kona (Fig. 1d). tially such prolonged, duration are thus a special case. The consequences of prolonged interaction 1.1. Kona with social, economic and physical infrastructure require special attention. Modern hazard mitiga- Our study area lies principally in the political tion campaigns tend to focus on building resil- districts of North and South Kona, collectively ience in advance rather than compensating retro- called Kona (Fig. 1d), which form the western spectively for loss and de¢cit (Omar and Alon, portion of Mauna Loa and of Hualalai shield 1994; Tobin, 1999; Van den Eyde and Veno, volcanoes. Both districts in Kona have been ad- 1999; Violanti et al., 2000). A valid question in versely impacted by lava £ows during historical Kona and elsewhere on the island is how can eruptions of Hualalai and Mauna Loa and by society best adapt to future eruptions of varying gas emissions from an on-going, 20-year old erup- durations? tion at K|lauea. The populations of North and South Kona in 2000 were 28 543 and 8589, respec- 1.3. Volcanoes tively (County of Hawaii, 2000, see table 1.5). North Kona includes the second largest city on Mauna Loa has erupted 39 times since 1832 the island (Kailua-Kona, population 9870; (Barnard, 1995), on average every 4.4 years. Sub- County of Hawaii, 2000, see table 1.7) and is aerial eruptions in 1851, 1859, 1916, 1919, 1926, more densely populated than South Kona. Risk 1949, and 1950 produced lava £ows in Kona (Fig. to society in Kona as a consequence of future 1c), and a submarine eruption occurred o¡shore volcanic activity at Mauna Loa and Hualalai of Kona in 1877 (Moore et al., 1985). The most has increased sharply since the last eruptions in recent eruption of Mauna Loa in Kona occurred the area, due to growth in population, infrastruc- during a 3-week period in 1950 and produced de- ture and its growing economic relationship with structive lava £ows (Macdonald, 1954; U.S. Geo- tourism. Most of Kailua is within 15 km of po- logical Survey, 2000). Mauna Loa last erupted in tential eruptive sources. South and North Kona 1984 and threatened Hilo, in the eastern portion are two of the three greatest areas of concern on of the island (Lockwood et al., 1987; Trusdell, the island identi¢ed by the state’s Lava Flow Haz- 1995). Hualalai erupted twice in Kona in ca. ard Mitigation Plan Technical Committee (Ha- 1800 (Kauahikaua et al., 2002), but the details waiki State Civil Defense, 2002). of these eruptions are less well documented than those for eruptions of Mauna Loa. Recurrence 1.2. Natural hazards and Kona intervals for Holocene time are on the order of 50 years (Moore et al., 1987). Hualalai has no Volcanic hazards may have a more prolonged summit caldera, unlike Mauna Loa, where signi¢-

VOLGEO 2711 6-1-04 182 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196

Fig. 1. Generalized maps of the island of Hawaiki. (a) Topography (contour interval = 1000 feet), volcanoes, major cities, and high schools surveyed (square = Kealakehe; circle = Konawaena). (b) Lava-£ow hazard zones. Bold lines outline volcanoes. (c) Dates of selected recent lava £ows; locations of main roads and highways. (d) Political districts. After U.S. Geological Sur- vey, 1997a.

cant volumes of lava can be impounded within the greatest hazard to property. Lava £ows occur the caldera. The current lava-£ow hazard-zone more frequently than ash fall in Kona, but they map (Wright et al., 1992) indicates that Hualalai are less frequent than gas emissions, which are is less hazardous than Mauna Loa and K|lauea, derived principally from K|lauea. The lava-£ow but future re¢nements to the hazard zone map at hazard zones (Fig. 1b) are ranked 1^9, with 1 Hualalai (Kauahikaua et al., 2002) may increase the most hazardous and 9 the least hazardous hazard levels at this volcano. (Wright et al., 1992). South Kona includes zones 1, 2 and 3. North Kona includes mostly zones 3 1.4. The nature of volcanic hazards in Kona and 4. All of Hualalai is currently in zone 4. Greater values for slope (Fig. 1a) and e¡usion Mullineaux and Peterson (1974), Mullineaux et rates compared to K|lauea contribute to faster- al. (1987), Heliker (1990), and Wright et al. (1992) moving lava £ows in Kona. The fastest lava described volcanic hazards on the island of Ha- £ow recorded in Hawaiki occurred in South waiki and produced hazard-zone maps for lava Kona during the 1950 eruption of Mauna Loa. £ows, tephra fall and volcanic gas emissions. Mul- One lava £ow traveled 24 km in 2.5 h, an average lineaux et al. (1987) suggested that lava £ows are rate of 9.6 km h31 (Macdonald, 1954; U.S. Geo-

VOLGEO 2711 6-1-04 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 183 logical Survey, 2000). The existence of faster £ows the belief that if an inhabited area is threatened during Hualalai’s last eruptions in ca. 1800 is de- by lava, evacuation will take place’. Furthermore, batable (Kauahikaua et al., 2002), but certainly they reported that, apart from evacuation, the steeper slopes enhanced local £ow-front advance most widely used of all other adjustments is the rates. Then and in the future, steep slopes and ‘bearing of losses when they occur’. Engineered high e¡usion rates combine to produce relatively measures are somewhat controversial and may fast-moving lava £ows, which pose heightened not be realistic options to mitigate lava-£ow haz- concern for society in Kona. ards in Kona (Mullineaux et al., 1987; Moore et Vog is dominated by aerosol particles in Kona al., 1987), where slopes are steep and travel times (U.S. Geological Survey, 1997b). Adverse e¡ects from source to ocean may be only a few hours or of vog in Kona include physical, chemical, and less, unless the measures are implemented well in biological damage to plants, water supplies, and advance of eruptions. Diversion of lava into oth- respiratory systems. On-going vog problems in erwise unthreatened areas is a particularly sensi- Kona are derived primarily from the 20-year-old tive issue. There are no current plans to imple- eruption of K|lauea, but problems also occurred ment engineered measures (e.g. diversion during the Mauna Ulu eruption of K|lauea in barriers) to protect against lava-£ow hazards in 1969^1974 (Mullineaux et al., 1987). Tephra fall Kona (Hawaiki State Civil Defense, 2002). from Hawaiian volcanoes is generally of greatest consequence near the eruptive sources and thins with distance from source. Tephra from explosive 2. The survey eruptions of Hualalai has not fallen in dense areas of Kailua-Kona and is generally con¢ned 2.1. Survey method to areas of less than about 10 km2 from the vents (Moore et al., 1987). Eruptions of Mauna Loa The questionnaire was designed for students, have not produced damaging amounts of ash in their parents or guardians, and adults at large. Kona. The questionnaire used in this study was modi¢ed Mullineaux et al. (1987) discussed issues sur- from one developed for studies of volcanic hazard rounding limitations of mitigating lava £ow com- risk perception in New Zealand (Ronan and pared to ash and gas hazards. In Hawaiki and Johnston, 2001). The questionnaire was reviewed elsewhere, rock walls have been hastily con- by scientists and emergency managers, pretested, structed to stop or retard the advance of lava and precoded. £ows, earthen dams have been constructed to re- Our data were derived from ¢ve separate tain lava, earthen diversion barriers and channels groups: 9th-grade students at (1) Kealakehe have been constructed and excavated to divert High School (n = 132) on Hualalai in North lava, bombs have been dropped to interrupt Kona and (2) Konawaena High School (n = 104) lava, and water has been sprayed on £ows to re- on Mauna Loa in South Kona (Fig. 1a); (3 and tard their movement (Jaggar, 1945; Macdonald, 4), parents or guardians of these students (n = 117 1958, 1962; Wentworth et al., 1961; Lockwood and 44), respectively; and (5) adults at large (de- and Torgerson, 1980; U.S. Army Corps of Engi- rived from a random survey of Post O⁄ce box neers, 1980; Colombrita, 1984; Barberi et al., holders (n = 51) and a survey of residents of 1993). More frequently though, people that have Kona awaiting £ights to Kona at the internation- been threatened by lava £ows in Hawaiki turn al airport in Honolulu (n = 14). We refer to these simply to evacuation, although other alternatives groups as North and South Kona students, North have involved appeals to the supernatural. Mur- and South Kona parents, and Adult-at-large. Re- ton and Shimabukuro (1974) reported that evac- turn rates were: students and airport samples uation is the only emergency adjustment with (100%), North Kona parents (89%), South Kona which Puna residents are familiar and that the parents (42%), and mail-out adults (10%). The warning and emergency system is ‘premised on district in which a student attends school is a

VOLGEO 2711 6-1-04 184 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 reasonable proxy for the district in which he/she most likely to erupt should be inversely related to resides. the repose time since the last signi¢cant eruption.

2.2. Survey purpose 2.3. Respondents (or sample)

The survey was designed to evaluate experience A total of 462 questionnaires were returned with hazard events and educational programs, (236 students, 226 adults). Demographic data understanding of the emergency warning system, for the student and adult samples, respectively, levels of hazard awareness, perceptions of vol- are: mean age (14.7 years, sd = 0.763 and 44.4 canic risk, and levels of preparedness in Kona, years, sd = 9.389); gender (male = 52.8% and and to identify potential underlying causes and 31.1%). Home and land ownership for adults in£uences. Hazard awareness or knowledge in£u- only is: 57.1% own their home, 31.6% rent, 19% ences risk perception, and perceptions are one own land in another town and 16% own a second driving mechanism for preparedness (Green et home. Ethnicity, on average, is: White 41.1%; al., 1981; Perry and Lindell, 1990; Perry and Hi- Hawaiian, part-Hawaiian and Paci¢c Islander rose, 1991; Lindell, 1994; Johnston et al., 1999). 36.5%; Japanese 16.7%; Filipino 12.2%; other Knowledge of a threat relates to salience of the 29.6%. Demographic characteristics, while not hazard, levels of past damage incurred, and the corresponding exactly to U.S. Census Bureau extent and timing of contact with hazard informa- (2000) data, do provide a representative pro¢le tion. Perceived risk has been linked to proximity of major ethnic groups and allow the tentative to the hazard source, perceived likelihood of fu- generalization of these data to the wider popula- ture disasters, and the perceived extent of impact, tion. as well as past experience in disasters. Hazard For the North and South Kona parent groups, adjustment re£ects perceived risk, the information about 75^76% were female (n = 119), compared to received regarding protective measures, extent of 53% for the Adult-at-large group. Female parents past damage, self-e⁄cacy (the individual’s percep- or guardians may be more inclined than their tion of their ability to play a role), and outcome male counterparts to participate in surveys admin- expectancy (the notion that a community can mit- istered through their children’s schools. More- igate hazard consequences; Bishop et al., 2002; over, 68% of single-parent families in Kona Lindell and Whitney, 2000; Paton, 2000). (24%) include female householders (County of The di¡erence in lapsed time since the last erup- Hawaii, 2000, see table 1.20). tions of Mauna Loa and Hualalai to a¡ect Kona provides an opportunity to compare threat 2.4. Knowledge of the volcanic hazard knowledge and risk perception as a function of the time since the eruptions. Potentially, percep- We explored threat knowledge through question of which volcano (Mauna Loa or Hualalai) is tions related to past perceptions of current and

Table 1 Perceptions that each volcano could erupt again by ¢ve survey groupsa N Kona St S Kona St S Kona P N Kona P Adult Total n = 131 n = 100 n =42 n = 116 n =65 n = 454 (%) (%) (%) (%) (%) (%) K|lauea 63.4 72.0 66.7 68.1 70.8 67.8 Mauna Loa 29.8 45.0 57.1 56.0 73.8 48.7 Hualalai 50.4 20.0 57.1 60.3 55.4 47.6 Mauna Kea 23.7 24.0 33.3 21.6 36.9 26.0 Kohala 10.7 15.0 2.4 6.9 7.7 9.5 None will erupt again 8.4 11.0 7.1 6.9 3.1 7.7 a N = North, S = South, St = Student, P = Parent, Adult = Adult-at-large.

VOLGEO 2711 6-1-04 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 185 future volcanic activity on the island, awareness K|lauea has been erupting continuously since of the most recent eruptions, understanding of the 1983, 26% would perceive that K|lauea could speed with which lava may £ow, experience in not erupt again. The data may re£ect a misper- hazard events, awareness of the lava-£ow hazard ception that, once K|lauea stops erupting, it could zone in which people live, and prior contact with not erupt again. hazard information. When asked, ‘Which two of four volcanoes are Knowledge of volcanism on the island is high most likely to erupt next’, Mauna Loa and Hua- but for Kona speci¢cally is low. For example, lalai were perceived to be more likely to erupt 89.6% (n = 387) of respondents recognized next than Mauna Kea and Kohala; this is a K|lauea as the volcano that has been erupting al- realistic perception. However, only 71% (n = 312) most continuously since 1983, although 10.4% selected Mauna Loa, and two-thirds (n = 289) se- (n = 45) were unaware of this fact. This lack of lected Hualalai. Nearly one-third of the respond- awareness is shared even by people who have re- ents (mostly from South Kohala District) selected sided in Kona for up to 43 years. Knowledge of Mauna Kea. These perceptions of which two vol- the threat posed to society in Kona as a conse- canoes are most likely to erupt is linked to prox- quence of future eruptions of Mauna Loa and imity to the volcano. Chi-square analyses suggest Hualalai is low. When asked, ‘Which volcanoes that perceptions of future activity at Mauna Loa do you think could erupt again’, about two-thirds and Hualalai are not proportional to the length of (68%) selected K|lauea, but only a minority be- time respondents have lived in Kona. lieved Mauna Loa and Hualalai could erupt again Perceptions of the timing of recent eruptions (Table 1). Furthermore, a surprising 8% (n = 35) also show limited understanding. Lava £ows em- indicated that ‘none will erupt again’. Chi-square placed during the latest eruptions of Hualalai in values indicate signi¢cant statistical di¡erences in ca. 1800 remain largely barren in Kona, and por- perceptions of future activity of Mauna Loa tions of the international airport and major high- (M2 = 39.485; P = 0.000) and Hualalai (M2 = ways are built on these £ows. When we asked, 41.616; P = 0.000). For the student groups, per- ‘When was the last time Hualalai volcano ception is related to proximity to the hazard erupted’, only one-third answered correctly (i.e. source (in this case the volcano on which each about 200 years ago (Table 2)). Less than one- student group attends school and presumably re- quarter answered correctly when we asked, sides; contrast Mauna Loa and Hualalai data in ‘When was the last time lava from Mauna Loa Table 1). Our data can be contrasted with the £owed into the ocean in Kona’ (i.e. about 50 ¢ndings of Murton and Shimabukuro (1974), years ago). The data show a sharp contrast be- who reported that 12% of the respondents in tween the responses of the student and adult sub- Puna District did not expect eruptions of K|lauea populations (Table 2). The di¡erences in aware- in the future (79% did) ^ a surprising ¢nding to ness between adults and students are surprising, us, given the recent and frequent activity of given the relative exposure of these two subpopu- K|lauea at that time. lations to hazard education programs, and sug- The low response for K|lauea in our sample is gests current programs perhaps do not focus on unlikely related to misinterpretation of our ques- Kona speci¢cally. Moreover, we found that, tion, because the ¢ve groups in Table 1 did not among those who indicated that they have been di¡er in their opinion (M2 = 2.285; P = 0.684). involved in a hazard education program or had Moreover, among the 90% who were aware of studied lava £ows in school, overall awareness of the long-term activity at K|lauea, 74.1% the eruptions was no higher than the total shown (n = 286) perceived that K|lauea could erupt in Table 2. This suggests that hazard education again, and chi-square analysis indicates that, may not have focused on the most recent activity again, the ¢ve groups did not di¡er in their per- in Kona. This low awareness of the most recent ceptions (M2 = 1.192, P = 0.879). It is not clear volcanic activity in Kona contrasts with the much why, among the subsample that knew that greater awareness of activity at K|lauea.

VOLGEO 2711 6-1-04 186 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196

Table 2 Awareness of years since last eruptions of Hualalai (about 200 years ago) and of Mauna Loa (in Kona, about 50 years ago) by student and adult subgroups Mauna Loa Hualalai Students Adults Total Students Adults Total n = 230 n = 212 n = 442 n = 229 n = 211 n = 440 (%) (%) (%) (%) (%) (%) About 500 years ago 8.7 9.0 8.8 4.8 3.8 4.3 About 200 years ago â â â 19.7 46.4 32.5 About 50 years ago 10.9 35.4 22.6 â â â About 20 years ago 4.3 7.5 5.9 7.4 2.8 5.2 Don’t know 76.1 48.1 62.7 68.1 46.9 58.0 Total 100 100 100 100 100 100 a Not applicable.

The potential high velocity of lava £ows in £ows, or ash fall had damaged their neighbor- Kona is a major issue for hazard mitigation. We hood. However, in a subsequent question, we asked, ‘In a future eruption, how quickly could a asked respondents to ‘List any damage that vog lava £ow reach the ocean in Kona’. The data in- has caused you or your family’s health or prop- dicate that about one-third (33.9%, n = 146) se- erty’. A slight majority of adults (52.3%, n = 111) lected the correct answer (i.e. 3 h or less), while and one-third (33.0%, n = 73) of students identi- slightly more (36.2%) selected 12 h. The remain- ¢ed impacts on health. These data could have der selected longer times of 3 days (5.8%, n = 25) implications for the nature of preparedness pro- and 1 month (24.1%, n = 104). Amongst those in- grams. Rather than focusing on hazard character- dicating involvement in a hazard education pro- istics per se, programs may be more e¡ective by gram or that had studied lava £ows in school, focusing on the interaction between these charac- overall awareness was unchanged (33%). We infer teristics and community needs. Very few respond- that respondents have a poor understanding of ents or their relatives have been threatened or the time scale with which lava £ows can travel, a¡ected by a damaging eruption. Only 36 of 422 and again, that education was not focused on respondents (8.5%) answered in the a⁄rmative. Kona’s situation. Hawaiian eruptions often allow relatively easy These data are particularly informative in re- access to observe passive lava £ows without ex- gard to preparedness. Most respondents live be- plosive activity and ash fall. Paton et al. (2001b) tween eruptive sources and the ocean, but it is not and Lindell and Perry (1992) found that only di- clear if people con£ate time to reach the ocean rect experience in hazard events was found to in- with time to reach them or to impact on them. £uence risk perception, as opposed to vicarious If the warning time exceeds that required to make experience (Sjo«berg, 2000). About one-third what they think are the necessary preparations, (37.7%, n = 170) indicated that they had seen ac- pre-eruption preparedness will decline. Pre-eruptive lava £ows. This is considerably less than 85% tion preparedness is even less likely for those re- of Puna residents who reported having experi- cording 1 month (24%). enced a number of eruptions on K|lauea (Murton Our data show that, on average, direct experi- and Shimabukuro, 1974). ence of natural hazard events that have damaged Knowledge of the lava-£ow hazard zones is at a the respondent’s neighborhood is low (28.3%, very low level. Lava-£ow hazard zones are de¢ned n = 122, missing data = 16) and very low for vol- for the entire island, i.e. all structures are located canic hazards. Of the 122 with experience, less within a zone or lie along the boundary of two or than 5.8% (n = 7) indicated that either vog, lava more zones. The salience of lava-£ow hazards

VOLGEO 2711 6-1-04 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 187 may be readily evaluated by the percentage of (31.2%, n = 72 compared to 14.8%, n = 28). For respondents who are aware of the zone in which ash fall, student and adult responses were similar they live. Only 15% (n = 44) of respondents indi- (average = 18.1%, n = 76). cated they knew the lava-£ow hazard zone in The U.S. Geological Survey (USGS) Hawaiian which they resided. In reality only 8 of these iden- Volcano Observatory (HVO) disseminates infor- ti¢ed a possible legitimate zone (13 did not re- mation through a weekly newspaper column titled spond). The data question utility of the zones Volcano Watch, which is published in local news- for residential property owners. Knowledge of papers and posted on the internet (since late the hazard zone is higher among those with prior 1994). To evaluate exposure to Volcano Watch, hazard experience and is in£uenced by ethnicity. which is devoted to topics such as volcanic haz- For example, Japanese respondents were about ards and processes on the island of Hawaiki and four times less likely to know the zone in which to informational topics related to other volcanoes they lived than the mean of other ethnicities in the U.S. and overseas, we asked, ‘Did you (17%). Knowledge also appears to be proportion- know that Volcano Watch is a Big Island news- al to the degree of lava-£ow hazard in which re- paper column on volcanoes’ to which on average spondents live. For example, 38.5% (n = 10) in 40% (n = 171) answered in the a⁄rmative. The hazard zones 2 and 3 were aware, 13.9% (n = 26) di¡erence in student and adult responses is statis- in zone 4 and 8.8% (n = 3) in zone 8. Sample sizes tically signi¢cant (32.1% versus 48.1%, M2 = were too small to test for statistical signi¢cance. 11.391; P = 0.001). Of those respondents who an- These data apply to the Adult-at-large and Kea- swered in the a⁄rmative, those who indicated lakehe groups. they read Volcano Watch last week included 41.7% (n = 101) of adults, but only 19.4% 2.5. Hazard education (n = 13) of students (average = 32.5%, n = 53). Sim- ilarly, 38.8% (n = 26) of students indicated they A slight majority (50.8%, n = 229, missing never read it compared to 0% for adults. On aver- data = 11) of respondents indicated they have age, 27% (n = 44) read Volcano Watch last month been involved with a hazard education program and a further 24.5% read it last year. The data that informed them about the types of hazards illustrate the contrast in use of di¡erent sources of included in this survey. Students claimed greater hazard information among students and adults. involvement (59.7%, n = 139) than adults (41.3%, This is not the same as utilizing it or even render- n = 90, M2 = 15.210, P = 0.000). 86.1% of students ing it meaningful in the manner anticipated. More (n = 118) selected school as the source of the in- work is needed to examine the factors that in£u- formation compared to 64.0% (n = 57) of adults. ence interpretation and how it is rendered salient. On average 30.1% (n = 68) selected Hawaiki Vol- People believe HVNP is a signi¢cant source of canoes National Park (HVNP), and 13.7% hazard education. A majority (52.3%) of student (n = 31) could not remember. However, only respondents indicated they have participated in a 5.1% (n = 7) of students selected other sources 3-day ¢eld trip to HVNP (mostly as 5th- or 6th- compared to 49.4% of adults (n = 44). The latter grade students), and about 80% (n = 168) of the data suggest that adults receive information from adults indicated they have visited HVNP and sources not recognized by students. learned about volcanic hazards. We did not eval- The extent to which lava-£ow, vog, and ash-fall uate what type of educational material the adults hazards have been taught or discussed at school is were exposed to at HVNP, but a question re- disproportional, with an emphasis on lava £ows. mains, what did they learn from displays and bro- A majority of students (53.7%, n = 124) indicated chures versus what was taught by HVNP per- they had studied or discussed lava £ows, com- sonnel or other o⁄cials? A concern is that pared to only about one-third of adults (34.9%, respondents may wrongly associate their mere n = 66). Similarly, more than twice as many stu- presence inside HVNP with hazard education, or dents had studied or discussed vog than adults confuse their knowing where information could

VOLGEO 2711 6-1-04 188 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 be obtained with their actual knowledge (Paton et 2.6. Perceptions of risk al., 2000). We asked respondents to rate the consistency of The public’s perception of risk has been linked all volcanic information that they had received in to proximity to the source of a hazard (i.e. Mauna Hawaiki. Adults (51%, n = 105) perceive informa- Loa and Hualalai), perceptions of the likelihood tion to be more consistent than students (30%, of future disasters, and the extent of impact, as n = 64), whereas students (43.2%, n = 92) are well as past experience in disasters. The island of more uncertain about the consistency than are Hawaiki is built of volcanoes, so society exists adults (25.2%, n = 52). This is surprising, since a everywhere on some portion of the ¢ve volcanoes. signi¢cantly greater percentage of students indi- In our survey, respondents live predominantly on cated that they had studied lava £ows and vog Hualalai and Mauna Loa, but also on the lower in school. The data suggest a need for clari¢cation £anks of Mauna Kea and Kohala. Respondents of some information, and whether this could re- were found to have a good understanding of the £ect other factors (e.g. independent search two volcanoes nearest to their home and identi¢ed through the web), which furnishes information Hualalai (74.7%, n = 339) and Mauna Loa (64.1%, that is less accurate or unrelated to the Hawaiian n = 291). Proximity of some respondents to situation. On average, 13.1% had not heard any- Mauna Kea moderated responses for Hualalai. thing and 12.2% indicated the information was To evaluate the popular perception of risk fairly to very inconsistent. across a variety of hazards that occur in Kona, To evaluate preferred sources of information, we asked participants, ‘Thinking about the chan- we asked participants where they would look for ces of property damage and loss of life or injuries, information about future volcanic activity on the which are the two most likely hazards that could island. Only one source (television and radio) is a¡ect you at home, school (students only) and favored by a majority of respondents (58.4%, work (adults only)’ (Fig. 2). Risk is generally per- n = 261); 34.5% (n = 154) selected USGS HVO ceived to be greatest at home, and lava £ows are website, a similar number selected Civil Defense, perceived to represent a greater threat than vog and 30.9% (n = 138) selected ‘internet’. Between and ash fall. The responses may re£ect the volume about 11% and 14% selected school, Police and of information received about each hazard as op- Fire Department, telephone book and local li- posed to an objective analysis of risk exposure brary. An interesting issue is the possibility that among all the hazards, although this cannot be respondents are recalling sources related to haz- determined from the data. ards with which they are familiar rather than de- To evaluate the perception of repose period, we scribing actual preferences (Ballantyne et al., asked, ‘When do you think the next hazard events 2000). The latter authors conducted a pre- and post-volcanic education survey. Information during the pre-survey revealed that pamphlets in the mail box were identi¢ed by respondents as their preferred medium to receive hazard information. However, the subsequent evaluation of a pam- phlet-based education strategy revealed that this medium proved highly unsuccessful. The authors inferred that people identi¢ed media as what they were familiar with rather than from thinking about the relationship between medium and e¡ectiveness. This is supported by the large num- Fig. 2. Perception of risk for hazards at home, school and ber of responses (937) to a question where re- work. Gas leak includes chemical spill; violence includes vio- spondents were not asked to make multiple selec- lent incident at school (students only) or at work (adults tions. only); storm includes with high winds and lightning.

VOLGEO 2711 6-1-04 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 189

measures that their family had taken. We then divided the measures into two groups, those that may also be used for frequent domestic emergencies (e.g. having a £ashlight, ¢rst aid kit, ¢re ex- tinguisher) and those applying only to infrequent hazard events (e.g. having the household water heater secured to the wall, buying extra insurance, having an emergency contact in another town, or having property inspected for vulnerability to natural hazards). The data do not represent preparedness measures taken to protect against a speci¢c hazard, but they do show the tendency for people to adopt measures that are less time-consuming Fig. 3. Average perception of time until the next hazard and less expensive to undertake (mean = 65.7%, event is likely to a¡ect respondent’s community. Gas leak in- sd = 16.0) rather than more involved and time- cludes chemical spill. consuming measures (26.5%, sd = 9.6; Mileti and O’Brien, 1992). For frequently occurring hazards, adults consistently indicated levels of prepared- are likely to a¡ect your community’? Results (Fig. ness higher than did students, but for infrequently 3) are shown for those perceiving that an event occurring hazards, students’ perceptions of pre- will occur ‘within the next year’ and ‘within the paredness were higher in all but two of the pre- next 1 to 10 years’. Responses for greater than 10 paredness measures. years are not shown but make up the residue. The The perception of preparedness of and within a results indicate that the times in which volcanic community can have important implications for hazards are perceived to likely a¡ect the respon- actual preparedness at the individual level. For dent’s community vary widely. Vog is considered example, Weinstein and Klein (1996) described to be the most likely hazard to a¡ect the respon- an ‘unrealistic optimistic bias’ in which a majority dent’s community within the next year, but ash of respondents rated themselves as being more fall and lava £ows are among the least likely. prepared (less vulnerable) and more skilful than For vog, the number of females (78%, n = 21) se- the average. While individuals may be aware of a lecting ‘within the next year’ was di¡erent from need for greater preparedness within the commu- the 44% (n = 10) of males. Data for vog are de- nity, the need is not recognized as applying to rived only from the Adult-at-large group. For them. As a result, individuals may be less likely lava £ows, the percentage of students who se- to participate in activities designed to increase lected within the next 1^10 years (31%, n = 71) preparedness and act on warning messages (John- was statistically di¡erent from the responses of ston et al., 1999). To evaluate the perception of adults (18.1%, n =39; M2 = 10.585, P = 0.005) and preparedness, we asked, ‘How prepared do you suggests a greater perception of risk among stu- believe the following people and groups of people dents. are for a damaging earthquake’, using a scale ranging from 1 (very prepared) to 4 (not at all 2.7. Preparedness prepared). On average, our data indicate that respondents perceive themselves as less prepared Data indicate that the majority of preparedness (mean 2.61, sd = 1.047) than their family (2.54, measures that have been adopted in Kona focus sd = 1.039) and emergency management agencies on frequent but low-impact household emergen- (2.07, sd = 1.369), but more prepared than the cies rather than infrequent but more damaging community (2.85, sd = 1.217) in general and imply natural hazards. From a list of 21 preparedness the presence of an unrealistic optimism. measures, we asked respondents to select those Mulilis and Duval (1995) and Lindell and

VOLGEO 2711 6-1-04 190 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196

Whitney (2000) found that a reduction in pre- occurrence of eruptions. The researchers also re- paredness results when persons assign responsibil- ported that many more people, especially old Ha- ity for their personal safety to others. To evaluate waiians, o¡er gifts to Pele. Lachman and Bonk the perception of responsibility for preparedness (1960) reported on belief in Pele during the 1960 for volcanic eruptions, we asked, ‘Who should be eruption in Puna, but suggested that belief was responsible for preparing the community for fu- not limited to any one religious creed, ethnicity, ture volcanic eruptions’ (Table 3). These data are age level, or degree of education. To evaluate the signi¢cant, in that respondents believe that indi- association of Pele with volcanic eruptions, we viduals in the community are less responsible than asked, ‘Who or what controls whether or not a the emergency management o⁄cials or agencies. Hawaiian volcano erupts’? For those selecting This comes as no surprise; considering that, in the only one answer (n = 376), on average a majority prior question, respondents perceived o⁄cials to (58.5%, n = 220) selected pressure beneath each be the most prepared for a damaging earthquake. volcano, whereas only 15% selected chance, The transfer of responsibility for oneself to others 12.8% Pele, 9% God, and 4% other. Students could re£ect di¡erences in expertise (Ballantyne et (20.8%), especially in South Kona, were more al., 2000) and the inference that those agencies likely to select Pele than adults (4.3%), whereas that are funded by tax dollars to manage hazards three-quarters of adults selected pressure beneath should take responsibility for the safety of people each volcano, compared to only 42% of students. in their community. 54 respondents chose to select multiple answers. We found that, on average, only one-third of Among these, 32 selected Pele, 31 God and 42 respondents indicated that their families had a pressure. Together, the data suggest a mixed per- plan of action for how to act in an emergency ception of the causes of volcanic eruptions in Ha- such as a lava £ow, hurricane, or earthquake. waiki, with beliefs rooted in physical volcanic pro- This suggests that plans involving lava hazards cesses held by a good majority (61%), followed by may be even less common. Furthermore, only spiritual beliefs in Pele and God by about one- one-third had practiced what to do during an third. Our ¢ndings show that belief in Pele is emergency at home; slightly more knew the loca- strongest among Spanish respondents. For exam- tion of an emergency evacuation shelter near their ple 44% (n = 7) of Spanish people selected Pele school or home. alone, Hawaiian (30%, n = 8), Filipino (22%, In the Puna District on K|lauea, Murton and n = 10), and part-Hawaiian (20%, n = 18). Less Shimabukuro (1974) found that a few Hawaiians than 13% of other ethnicities such as White, Jap- believed that a kahuna (priest) or Madam Pele anese, and Portuguese selected Pele alone. The (the volcano goddess) will tell them when an erup- data suggest that belief in Pele as a controlling tion will occur. In contrast, they reported that factor in volcanic eruptions is shared by a minor- many haole (Caucasian) newcomers believed that ity and agree with the ¢nding of Lachman and scienti¢c monitoring by the HVO can predict the Bonk (1960) that belief in Pele is not limited to

Table 3 Perception of responsibility for preparing the community for future volcanic eruptions by student and adult subgroups Students Adults Total M2 ; P a n = 230 n = 212 n = 442 (%) (%) (%) Civil Defense 78.7 89.2 83.7 8.843; 0.003 USGS, HVO 56.5 74.1 64.9 14.896; 0.000 Police, Fire departments 57.8 62.7 60.2 1.110; 0.292 Every individual in the community 49.1 58.5 53.6 3.886; 0.049 a M2 = Chi-square value; P = probability. Missing data = 20. USGS = U.S. Geological Survey. HVO = Hawaiian Volcano Obser- vatory.

VOLGEO 2711 6-1-04 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 191 any one ethnic group. These data have signi¢cant been a¡ected by a damaging eruption. It is possi- implications for risk perception, preparedness, ble that widespread direct exposure to non-dam- and attentiveness to public education. aging lava £ows (37%) has fostered the perception that future volcanic eruptions will be similar to those they have witnessed on television or in per- 3. Discussion son (e.g. at K|lauea) and thus caused a normal- ization bias (Mileti and O’Brien, 1993; Johnston 3.1. Awareness and preparedness et al., 1999), rendering respondents less attentive to hazard information and less likely to respond Overall, our data suggest that knowledge of the to warnings than they might be otherwise. threat to Kona from eruptions of Mauna Loa and With these levels of awareness and knowledge, Hualalai is low. Why is this so? Lack of exposure preparedness is predictably at relatively low levels. to hazard programs may not be the cause ^ a The broader issue of hazard preparedness (or a majority of respondents indicated they had par- lack thereof) is not speci¢c to the volcanic threat. ticipated in hazard education programs. The data Moreover, individuals su¡er from an unrealistic suggest a need for analysis of how their pre-exist- optimism bias. Because respondents also believe ing beliefs and expectations impinge on informa- that individuals in the community are less respon- tion received (e.g. transfer of responsibility). Sur- sible than the emergency management o⁄cials or prisingly, respondents who participated in a agencies, they may be less inclined to attend to hazard education program or studied lava-£ow hazard information or react to warnings. In other hazards in school did not show improved aware- words, the very fact that speci¢c agencies (e.g. ness of the most recent eruptions or the velocity scienti¢c, emergency management) disseminate in- with which lava £ows can travel compared to formation and advice on preparedness may ac- those who had not participated in education pro- tually reduce the perceived need for personal pre- grams and did not study lava £ows in school. paredness as a consequence of a transfer of Perhaps the information provided may not have responsibility (Mulilis and Duval, 1995; Ballan- focused on Kona’s situation, or it was not re- tyne et al., 2000). tained by the respondents, because exposure has There is a need for education programs that not greatly improved their knowledge or under- target speci¢c groups and identify methods that standing of their volcanic hazards. This suggests a promote information exchange between students need to explore the content of education pro- and adults. Media like Volcano Watch are read grams received prior to the 9th grade and their by the adult population (especially Whites) but e¡ects on improving awareness and understand- not by students. In both cases, questions regarding. ing the meaning and utilization of this source for Within the Kona hazardscape, volcanic hazards preparedness should be considered. Identifying are given a low priority by residents. Knowledge primary preferences for accessing volcanic infor- of the lava hazard zones is at a very low level. mation is an important consideration. Of eight Lava £ows and ash fall were perceived to be the potential sources, only one (television or radio) least two hazards that are likely to a¡ect the re- was stated as being preferred by a majority of spondents’ community within the next 10 years. respondents. However, this may re£ect familiarity This is potentially problematic given the volca- rather than an objective assessment of the me- noes’ potential for sudden onset and long dura- dium. tion. On average, direct experience in a damaging 3.2. Engineered measures natural hazard of any type is low and is much lower for volcanic hazards, although experience Engineered measures to mitigate lava-£ow haz- with vog appears to be much higher. Moreover, ards are not planned for Kona and are unlikely to very few respondents have relatives who have be implemented (Hawaiki State Civil Defense,

VOLGEO 2711 6-1-04 192 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196

2002). Nonetheless, if engineered solutions are im- 3.4. North Kona versus South Kona plemented, but cannot be guaranteed 100% e¡ective, additional outreach is required to counter Perceptions of past and future volcanic activity phenomena (i.e. risk homeostasis) that may in- are linked to proximity to the hazard source, es- crease risk-taking behavior. If engineered solu- pecially among students. For example, students tions are not implemented, then additional com- from each school district are more knowledgeable munity outreach is needed to increase awareness of, and perceive a greater threat from, the volcano of the uniqueness of the volcanic threat in Kona on which they attend school and presumably live. and to promote adoption of realistic preparedness Similarly, parents from each school are more and response measures that will foster resilience in knowledgeable of the recent volcanic activity at future eruptions. this volcano. Parents from both schools perceive Mauna Loa and Hualalai as the two volcanoes 3.3. Adult versus student most likely to erupt next, although Hualalai is perceived to be less likely to erupt by South We chose to focus on the contrast between stu- Kona parents compared to North Kona parents. dent and adult populations, because there is a A greater percentage of parents from South Kona growing tendency for hazard education programs (one-third) perceive damaging lava £ows within to target children. This appears to be based on the the next 10 years as opposed to North Kona pa- ease with which children can be approached in rents (11.5%). This time frame may reduce the large numbers (e.g. in school settings) but also perceived urgency to prepare. A similar, but less on an assumption that children will later share pronounced di¡erence was observed between the the information with family members, and those students from these districts and together suggests with neighbors, and so on. This assumption needs a greater perception of threat from future lava validating. Our data show that student and adult £ows in the area of higher hazard (i.e. Mauna perceptions di¡er markedly in many instances. Loa in South Kona). Overall awareness of the Our data show that students have received most recent lava £ows at both volcanoes remains more hazard education in general, as well as spe- low. Nonetheless, awareness of the most recent ci¢c to volcanic hazards, yet students show lower activity is greater for Hualalai than for Mauna awareness of the most recent activity of their Loa in both North and South Kona, and may neighboring volcanoes and the fact that those vol- be a re£ection of the greater salience of Hualalai canoes could erupt again. Moreover, fewer stu- as a volcano or recent outreach e¡orts such as dents reported that volcanic information was con- USGS’s Volcano Watch to increase public aware- sistent, and a greater number were uncertain ness of Hualalai as a volcano in repose that will about the consistency of the volcanic information. likely erupt again. However, despite this, students were more likely than adults (by a factor of 1.7) to perceive that a 3.5. In£uence of ethnicity lava £ow could a¡ect their community within the next 10 years. Ethnicity was found to in£uence the beliefs As anticipated, experience in hazardous events about whom or what controls volcanic eruptions, is directly related to age. Surprisingly though, perceptions of future volcanism in Kona, the ex- adults consistently indicated that their family tent to which respondents have received hazard had taken more household preparedness measures education, and perceptions of who is responsible for frequent emergencies, whereas students re- for preparing communities for future eruptions ported higher levels of more involved measures and preparedness. such as those needed for more infrequently occur- Our data suggest that belief in Pele as the sole ring hazards. This is an interesting inconsistency controlling factor of Hawaiian eruptions is low in that requires further analysis to better under- Kona overall, ranging between 20% and 44% for stand. people of Part Hawaiian, Filipino, Hawaiian, and

VOLGEO 2711 6-1-04 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 193

Spanish ethnicities, respectively; whereas such be- information contained in Volcano Watch is less liefs were below 13% for other ethnicities. In con- likely to be disseminated within the Japanese trast, belief in pressure alone as the controlling community. Together these data suggest that factor was shared by a majority to 71% of people awareness of the volcanic threat in Kona varies of Hispanic (not Spanish or Portuguese), Japa- signi¢cantly among ethnic groups, but that partic- nese, and White ethnicities. A modern study of ipation in hazard education programs is not a Hawaiian culture and spiritual beliefs regarding prerequisite for awareness. The data also suggest volcanism and its in£uence on decision making, some de¢ciencies in the content of hazard educa- preparedness, and mitigation is long overdue. tional material, the ability for respondents to re- We found that Whites are more aware of the call information, or widespread dissemination and volcanic threat from Hualalai and Mauna Loa acceptance of educational material. Education in- than are respondents of other ethnicities. How- itiatives that are tailored to meet the needs of ever, a minority of Whites (41.8%) indicated speci¢c parts of communities are needed. In other they had participated in a hazard education pro- words, the blanket provision of the same informa- gram, compared to 57% of other ethnicities, and tion is unlikely to be e¡ective and its failure to suggests that factors other than involvement in accommodate speci¢c needs may engender a sense hazard education programs in£uence awareness of mistrust in the agencies providing information. and perceptions of volcanic activity. Paci¢c Is- landers (excluding Hawaiian) were much more likely than other ethnicities to perceive USGS as 4. Conclusions the party responsible for preparing the community for future eruptions (94% compared to 64% for At Kona (and throughout the Big Island) aware- other ethnicities) and less than half as likely to ness of volcanism is extremely high, but this gen- perceive responsibility to lie with individuals eral awareness has not translated into knowledge (24% compared to 55% for other ethnicities). Sim- of the speci¢c future threat to Kona communities. ilarly, Japanese respondents (95%) were more The low levels of individual preparedness for vol- likely than other ethnicities (82%) to attribute re- canic eruptions in Kona is consistent with both the sponsibility for community preparedness for vol- low to moderate knowledge of the threat posed by canic eruptions to Civil Defense. These observa- Mauna Loa and Hualalai and the ¢ndings of tions highlight the importance of examining how others that, historically, few adjustments for lava- cultural and ethnic factors in£uence interaction £ow hazards have been made in Hawaiki. Kates with hazard issues and agencies. We infer that (1971) discussed the low levels of preparedness these individuals have transferred responsibility that can be expected in communities where the for preparedness to the authorities and that these frequency of a hazard is low. groups may be less inclined to attend to informa- The greater frequency of eruptions from tion and respond to warnings. Filipino, Spanish, Mauna Loa in South Kona suggests a need to and Japanese Americans were much less likely to improve community preparedness for volcanic perceive Mauna Loa as a volcano that could eruptions there; but a greater number and density erupt than were other ethnicities and so may be of people and infrastructure occur downslope of less likely to make preparations for future erup- potential eruptive sources in North Kona. The tions. Portuguese and Spanish respondents (80% majority of respondents are not likely to make and 73%) were found to have participated in haz- plans for an eruption, as they do not believe an ard education programs much more than other eruption will occur. Furthermore, even for those ethnicities (about 50%), yet Spanish respondents who believe an eruption could or is likely to oc- showed lower awareness of the volcanic threat cur, there is no indication that this perception has from Mauna Loa. Finally, only 21% of Japanese translated into increased preparedness such as respondents were aware of Volcano Watch com- having a response plan. This re£ects the antici- pared to 43% of other ethnicities, suggesting that pated timing of hazard activity and speed of onset

VOLGEO 2711 6-1-04 194 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196

^ these combine to reduce the perceived urgency education programs, because Paton et al. about preparedness. An expectation is that a long (2001a) found that people are more likely to listen warning period will provide su⁄cient time for to hazard-related information and make adjust- preparation (i.e. between onset of activity and it ments to behavior when the information has an reaching them). Such perceptions may be di⁄cult immediate e¡ect on their livelihood or well-being. obstacles in encouraging new ways of adjusting to Future hazard education initiatives must also the volcanic threat in Kona. One thing is clear evaluate the changes in threat knowledge, risk though: current community understanding and perception, and preparedness as a result of expo- preparedness on Hualalai, and particularly sure to the educational material. Achieving resil- Mauna Loa falls short of that required for a vol- ience requires that attention is directed towards canic crisis, particularly for those eruptions with the nature of the relationship between scienti¢c short onset and high e¡usion rates on steep slopes and emergency management agencies and com- that would impact Kona in just a few hours (and munities, with the emphasis on community devel- some sparsely inhabited areas in less time than opment and empowerment (Paton, 2000). A valid that). Misconceptions about potential future erup- question in Kona and elsewhere on the island is tions from these volcanoes, and realistic times for how can we persuade society to make the adapta- lava £ows to reach developed areas, contribute to tions needed for both long- and short-duration perceptions of low risk and need to be corrected. eruptions from sites other than the currently ac- Scienti¢c and technical knowledge in itself is tive K|lauea east rift zone. This calls for strategies likely to be ine¡ective as a means of facilitating that involve more than just the provision of fac- preparedness. This is not to say that it should not tual information (Paton, 2000). Creating a realis- be provided. Indeed it should. However, it is vi- tic understanding of the likelihood of future erup- tally important, ¢rstly, to identify and correct tions is an important starting point to facilitate gaps or misconceptions in people’s knowledge. this shift of thinking. Secondly, information should be presented in a manner that is consistent with the needs and expectations of the community and in a way that is Acknowledgements (1) consistent with the reasoning processes involved in rendering information meaningful and Funding for this project was provided by the (2) readily used in preparedness actions. Our ad- Jaggar Bequest Fund; the Center for the Study vanced knowledge of the physics of volcanic erup- of Active Volcanoes at the University of Hawaiki; tions must be integrated with understanding how the USGS Volcano Hazards Program; and the community members di¡er in regard to how they Hawaiki Sea Grant College Program. We thank obtain, process, and utilize information. This is Principal William Murakami and Mr. Jim Wiley vital in the process of promoting and sustaining at Kealakehe and Konawaena High Schools, and community resilience and the ability of individu- the teachers and administrative sta¡ of these als to adapt to the consequences of future erup- schools, for providing access to students. Re- tions in Kona. spondents deserve a special thanks. Reviews and These ¢ndings can be used to focus future haz- comments to the pilot questionnaire by Frank ard education initiatives or surveys. Education Trusdell and Jim Kauahikaua are appreciated. initiatives that promote general hazard awareness and focus on the situation and needs at county or state level are only the ¢rst steps in building community resilience to natural hazards. Initiatives References are needed now that focus speci¢cally on Kona, Adams, J., 1995. Risk. UCL Press, London, 228 pp. its environment and the characteristics of its con- Ballantyne, M., Paton, D., Johnston, D., Kozuch, M., Daly, stituent groups. The high reports of damage from M., 2000. Information on volcanic and earthquake hazards: vog may be e¡ective starting points for future the impact on awareness and preparation. Institute of Geo-

VOLGEO 2711 6-1-04 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196 195

logical and Nuclear Sciences Science Report 2000/2, pp. 1^ parative shifts in knowledge and risk. Disaster Prev. Man- 45. age. 8, 118^126. Barberi, F., Carapezzam, M.L., Valenza, M., Villari, L., 1993. Kartez, D., 1982. Emergency planning implications of local The control of lava £ow during the 1991-92 eruption of Mt. governments’ responses to Mount St Helens. Nat. Hazard Etna. J. Volcanol. Geotherm. Res. 56, 1^34. Res. Working Pap. 46, 1^37. Barnard, W.M., 1995. Mauna Loa Volcano: Historical erup- Kates, R.W., 1971. Natural hazard in human ecological per- tions, exploration, and observations (1779-1910). In: Rho- spective: hypotheses and model. Econ. Geogr. 47, 438^451. des, J.M., Lockwood, J.P. (Eds.), Mauna Loa Revealed: Kauahikaua, J., Cashman, K.V., Clague, D.A., Champion, D., Structure, Composition, History, and Hazards. Geophys. Hagstrum, J.T., 2002. Emplacement of the most recent lava Monogr. 92, pp. 1^19. £ows on Hualalai Volcano, Hawaii. Bull. Volcanol. 64, 229^ Bishop, B., Paton, D., Syme, G., Nancarrow, B., 2002. Coping 253. with environmental degradation: Salination as a community Lachman, R., Bonk, W.J., 1960. Behavior and beliefs during stressor. Network 12, 1^15. the recent volcanic eruption at Kapoho Hawaii. Science 131, Burby, R.J., 1998. Co-operating with Nature: Confronting 1095^1096. Natural Hazards with Land-use Planning for Sustainable Lindell, M.K., 1994. Perceived characteristics of environmental Communities. Joseph Henry Press, Washington, DC, 356 hazards. Int. J. Mass Emerg. Disasters 12, 303^326. pp. Lindell, M., Perry, R.W., 1992. Behavioral Foundations of Colombrita, R., 1984. Methodology for the construction of Community Emergency Planning. Hemisphere Publishing, earthen barriers to divert lava £ows: the Mt. Etna 1983 New York, 309 pp. eruption. Bull. Volcanol. 47, 1009^1038. Lindell, M.K., Whitney, D.J., 2000. Correlates of household County of Hawaii, 2000. Hawaii County Data Book. Hawaii seismic hazard adjustment adoption. Risk Anal. 20, 13^15. County Department of Research and Development, URL: Lockwood, J.P., Torgerson, F.A., 1980. Diversion of lava http://www.hawaii-county.com/databook_current/sec- £ows by aerial bombing-lessons from Mauna Loa Volcano, tion01.htm. Hawaii. Bull. Volcanol. 43, 721^741. D’Ercole, R., Rancon, J.P., Lesales, T., 1995. Living close to Lockwood, J.P., Lipman, P.W., 1987. Holocene eruptive his- an active volcano: popular hazard perception in three com- tory of Mauna Loa Volcano. In: Decker, R.W., Wright, munities of North Martinique (FWI). IUGG XXI Gen. As- T.L., Stau¡er, P.H. (Eds.), Volcanism in Hawaii. U.S. sembly Abstracts, B20. Geol. Surv. Prof. Pap. 1350, 509^536. Fletcher III, C.H., Grossman, E.E., Richmond, B.M., Gibbs, Lockwood, J.P., Dvorak, J.J., English, T.T., Koyanagi, R.Y., A.E., 2002. Atlas of Natural Hazards in the Hawaiian Okamura, A.T., Summers, M.L., Tanigawa, W.R., 1987. Coastal Zone. U.S. Geol. Surv. Geol. Invest. Ser. Map I- Mauna Loa 1974-1984: A decade of intrusive and extrusive 2761, pp. 1^182. activity. In: Decker, R.W., Wright, T.L., Stau¡er, P.H. Green, M., Perry, R.W., Lindell, M., 1981. The March 1980 (Eds.), Volcanism in Hawaii. U.S. Geol. Surv. Prof. Pap. eruptions of Mount St Helens: Citizen perceptions of volca- 1350, 537^570. no hazard. Disasters 5, 49^66. Macdonald, G.A., 1954. Activity of Hawaiian volcanoes dur- Hamnett, M.P., et al., 1993. Hawaii Coastal Hazard Mitiga- ing the years 1940-1950. Bull. Volcanol. 15, 119^179. tion Planning Project, Phase I: Findings, Recommendations Macdonald, G.A., 1958. Barriers to protect Hilo from lava and Technical Documents. O⁄ce of State Planning, Hono- £ows. Pac. Sci. 12, 258^277. lulu, HI, 239 pp. Macdonald, G.A., 1962. The 1959 and 1960 eruptions of Hamnett, M.P., Davidson-Oh, K., Bay, M., Bay, J., Bryant, K|lauea volcano, Hawaii and the construction of walls K., 1996. Hawaii Coastal Hazard Mitigation Planning Proj- to restrict the spread of lava £ows. Bull. Volcanol. 24, ect, Phase II Final Report. Findings, Recommendations and 250^294. White Papers. Social Science Research Institute, University Mileti, D.S., Sorensen, J.H., 1990. Communication of Emer- of Hawaii, Honolulu, HI, 93 pp. gency Public Warnings: A Social Science Perspective and Hawaiki State Civil Defense, 2002. Lava £ow hazard mitiga- State-of-the-Art Assessment. ORNL-6609, Oak Ridge Na- tion plan: reducing the risk of lava £ows to life and prop- tional Laboratory, Oak Ridge, TN. erty. Hawaii State Civil Defense Report 11/2002, pp. 1^75. Mileti, D.S., Fitzpatrick, C., 1992. The casual sequence of risk Heliker, C.C., 1990. Volcanic and Seismic Hazards on the communication in the Park¢eld earthquake prediction ex- Island of Hawaii. U.S. Geol. Surv. Gen. Interest Publ. 48. periment. Risk Anal. 9, 20^28. Jaggar, T.A., 1945. Volcanoes Declare War. Logistics and Mileti, D.S., O’Brien, P., 1992. Warnings during disaster: nor- Strategy of Paci¢c Volcano Science. Paci¢c Paradise Ltd., malizing communicated risk. Soc. Probl. 39, 40^57. Honolulu, HI, 16 pp. Mileti, D.S., O’Brien, P., 1993. Public response to aftershock Johnston, D.M., Houghton, B.F., 1995. Secondary school warnings. U.S. Geol. Surv. Prof. Pap. 1553-B, 31^42. children’s perceptions of natural hazards in the North Is- Moore, J.G., Fornari, D.J., Clague, D.A., 1985. Basalts from land, New Zealand. N.Z. J. Geogr. 99, 18^26. the 1877 submarine eruption of Mauna Loa, Hawaii, new Johnston, D.M., Bebbington, M.S., Lai, C.D., Houghton, data on the variation of palagonization rate with temper- B.F., Paton, D., 1999. Volcanic hazard perceptions: com- ature. U.S. Geol. Surv. Bull. 1663, 11.

VOLGEO 2711 6-1-04 196 C.E. Gregg et al. / Journal of Volcanology and Geothermal Research 130 (2004) 179^196

Moore, R.B., Clague, D.A., Meyer, R., Bohrson, W.A., 1987. Perry, R.W., Hirose, H., 1991. Volcano Management in the Hualalai Volcano: A preliminary summary of geologic, pet- United States and Japan. JAI Press, Greenwich, CT, 230 pp. rologic and geophysical data. In: Decker, R.W., Wright, Ronan, K.R., Johnston, D.M., 2001. Correlates of hazard ed- T.L., Stau¡er, P.H. (Eds.), Volcanism in Hawaii. U.S. ucation programs for youth. Risk Anal. 21, 1055^1063. Geol. Surv. Prof. Pap. 1350, 571^585. Saarinen, T.F., Sell, J.L., 1985. Warning and Response to Mt Mulilis, J., Duval, T.S., 1995. Negative threat appeals and St. Helen’s Eruption. State University of New York Press, earthquake preparedness: a person-relative-to-event (PrE) Albany, NY, pp. 201^205. model of coping with threat. J. Appl. Soc. Psychol. 23, Sjo«berg, L., 2000. Factors in risk perception. Risk Anal. 20, 1^ 1319^1339. 11. Mullineaux, D.R., Peterson, D.W., 1974. Volcanic hazards on Tobin, G.A., 1999. Community resilience: The holy grail of the island of Hawaii. U.S. Geol. Surv. Open File Report 74- hazards planning. Environ. Hazards 1, 13^26. 239, 1^61, two folded maps. Trusdell, F.A., 1995. Lava £ow hazards and risk assessment on Mullineaux, D.R., Peterson, D.W., Crandell, D.R., 1987. Vol- Mauna Loa volcano Hawaii. In: Rhodes, J.M., Lockwood, canic hazards in the Hawaiian Islands. In: Decker, R.W., J.P. (Eds.), Mauna Loa Revealed: Structure, Composition, Wright, T.L., Stau¡er, P.H. (Eds.), Volcanism in Hawaii. History, and Hazards. Geophys. Mon. 92, pp. 327^336. U.S. Geol. Surv. Prof. Pap. 1350, 599^621. U.S. Army Corps of Engineers, 1980. Lava £ow control, Is- Murton, B.J., Shimabukuro, S., 1974. Human adjustment to land of Hawaii. Review report and environmental impact volcanic hazard in Puna District Hawaii. In: White, G.F. statement. U.S. Army Corps of Engineers, Honolulu Dis- (Ed.), Natural Hazards: Local, National, Global. Oxford trict, 116, appendices. University Press, New York, pp. 151^159. U.S. Census Bureau, 2000. American Fact Finder Quick Ta- Nathe, S., Gori, P., Greene, M., Lemersal, E., Mileti, D.S., bles. URL: http://fact¢nder.census.gov/bf/. 1999. Public education for earthquake hazards. Natural U.S. Geological Survey, 1997. Volcanic and Seismic Hazards Hazards Informer URL: http://www.colorado.edu/hazards/ on the Island of Hawaii. U.S. Gov. Printing O⁄ce, 48 pp. informer/infrmr2/infrm2wb.htm. U.S. Geological Survey, 1997. Volcanic Air Pollution - A Haz- Omar, H., Alon, N., 1994. The continuity principle: A uni¢ed ard in Hawaii. U.S. Geol. Surv. Fact Sheet 169-97, p. 1. approach to disaster and trauma. Am. J. Commun. Psychol. U.S. Geological Survey, 2000. 50th Anniversary of Mauna 22, 273^287. Loa’s Most Spectacular Eruption. URL: http://hvo.wr. Paton, D., 2000. Emergency planning: Integrating community usgs.gov/maunaloa/history/50_06_01/. development, community resilience and hazard mitigation. Van den Eyde, J., Veno, A., 1999. Coping with disastrous J. Am. Soc. Prof. Emerg. Manag. 7, 109^118. events: An empowerment model of community healing. Paton, D., Millar, M., Johnston, D., 2001a. Community resil- In: Gist, R., Lubin, B. (Eds.), Response to Disaster. Taylor ience to volcanic hazard consequences. Nat. Hazards 24, and Francis, Philadelphia, PA. 157^169. Violanti, J.M., Paton, D., Dunning, C., 2000. Posttraumatic Paton, D., Johnston, D., Bebbington, M., Lai, C.D., Hough- Stress Intervention: Challenges, Issues and Perspectives. ton, B.F., 2001b. Direct and vicarious experience of volcanic Charles C. Thomas, Spring¢eld, IL. hazards: implications for risk perception and adjustment Weinstein, N.D., Klein, W.M., 1996. Unrealistic optimism: adoption. Aust. J. Emerg. Manag. 15, 58^63. present and future. J. Soc. Clin. Psychol. 15, 1^8. Paton, D., Smith, L.M., Johnston, D.J., 2000. Volcanic haz- Wentworth, C.K., Powers, H.A., Eaton, J.P., 1961. Feasibility ards: Risk perception and preparedness. N.Z. J. Psychol. 29, of a Lava-diverting Barrier at Hilo, Hawaii. Pac. Sci. 15, 84^88. 352^357. Perry, R.W., 1990. Volcanic hazard perceptions at Mount Wright, T.L., et al., 1992. Map showing lava-£ow hazard Shasta. Environ. Prof. 12, 312^318. zones, Island of Hawaii. U.S. Geol. Surv. Misc. Field Stud- Perry, R.W., Lindell, M.K., 1990. Living with Mount St. Hel- ies Map MF-2193, Scale 1:250,000, one folded map. ens: Human Adjustment to Volcano Hazards. Washington Yoshii, H., 1992. Disaster warnings and social response - erup- University Press, Washington, DC, 205 pp. tion of Mount Unzen in Japan. Disaster Manag. 4, 207^214.

VOLGEO 2711 6-1-04