Reducing risk from lahar hazards: concepts, case studies, and roles for scientists

Case study 1. When economics and politics trump science

Following the June 15, 1991, eruption of Mount Pinatubo (Philippines), lahars and volcanic fluvial sedimentation threatened many downstream communities. Geologists from a number of institutions met with officials at local, provincial, and national levels to explain the threats and to evaluate and discuss proposed countermeasures. Due to political pressures (Rodolfo[1995]), officials ultimately adopted a lahar mitigation strategy that was based on the construction of parallel containment dikes close to the existing river channels, using easily erodible fresh sand and gravel deposits of earlier lahars as the construction material. Appropriation of the private land needed for lahar containment areas of adequate size was viewed by officials as too politically costly. Officials hoped the dikes would divert lahars and floods past vulnerable communities. However, nearly all the geologists involved in the discussions expressed the opinion that this was a poor strategy because (a) channel gradients were too low for effective sediment conveyance and deposition would occur in the wrong places, (b) dike placement did not provide adequate storage capacity and dikes would be overtopped or breached, (c) most of the dikes were not revetted and would be easily eroded by future lahars, and (d) people would be lured back to live in still-dangerous hazard zones. The advice of the scientists was not heeded, and over the next several years many of these predictions came true, including breached dikes due to lahar erosion and overtopped dikes due to sediment infill. Lahars breaking through the levees caused fatalities and destroyed many homes. A government official later explained (to TCP) that political considerations prompted the decisions to minimize the area of condemned land and build lahar catch basins that were too small. He felt that the plan recommended by the geologists would have angered too many people and that it was better for officials to be seen doing something rather than nothing, even if the chance of success was low. Indeed, political and economic forces can override scientific recommendations (Tayag and Punongbayan[1994]; Rodolfo[1995]; Janda et al.[1996]; Newhall and Punongbayan[1996]; Crittenden[2001]).

Slope stabilization and erosion control

Volcanic ash mantling hillslopes is extremely vulnerable to rapid surface erosion and shallow landsliding, and it is easily mobilized as lahars by heavy rain (e.g., Collins and Dunne [1986]; Pierson et al. [2013]). Even after long periods of consolidation and revegetation, ash-covered slopes can fail on massive scales and result in catastrophic lahars (Scott et al. [2001]; Guadagno and Revellino [2005]). Various methods of slope stabilization, slope protection, and erosion control can limit shallow landsliding or surface erosion in disturbed landscapes that could produce extreme sediment inputs to rivers (Figure 5), although most of these approaches are intensive, costly, and generally limited to hillside-scale problem areas (see overviews in Theissen [1992]; Morgan and Rickson [1995]; Gray and Sotir [1996]; Holtz and Schuster [1996]; Schiechtl and Stern [1996]; Beyers [2004]; Valentin et al. [2005]). These are only briefly summarized here. Options for drainage-basin-scale slope stabilization and erosion control are more limited, have been tested mostly in basins disturbed by wildfire rather than by volcanic eruptions, and are not always effective (Beyers [2004]; deWolfe et al. [2008]).

Regardless of scale of application, slope stabilization and erosion control techniques attempt to either (a) prevent shallow landsliding by mechanically increasing the internal or external forces resisting downslope movement, decreasing the forces tending to drive downslope movement, or both; or (b) prevent rapid surface erosion and sediment mobilization on slope surfaces and in rills, gullies, and stream channels (Gray and Sotir [1996]; Holtz and Schuster [1996]). Inert materials used to stabilize slopes and control erosion include steel, reinforced concrete (pre-cast elements or poured-in-place), masonry, rock, synthetic polymers, and wood, although many of these degrade and weaken with time. Biotechnical stabilization (Morgan and Rickson [1995]; Gray and Sotir [1996]) uses live vegetation to enhance and extend the effectiveness of many engineered structures.

Forces resisting slope failure or erosion can be maintained or augmented by a variety of approaches (Morgan and Rickson [1995]; Gray and Sotir [1996]; Holtz and Schuster [1996]). Counterweight fills, toe berms, retaining walls, and reinforced earth structures can buttress toes of slopes. To maintain buttressing at a toe slope, revetments using riprap, gabion mattresses, concrete facings, and articulated block systems can prevent toe-slope erosion. Anchors, geogrids (typically wire-mesh mats buried at vertical intervals in a slope face), cellular confinement systems consisting of backfilled three-dimensional structural frameworks; micro-piles, deeply rooted woody vegetation, chemical soil binders, and drains to decrease internal pore pressures can increase the shear strength of natural or artificial slopes. To reduce the driving forces, proven methods include regrading to lower slope angles, and weight reduction of structures or materials placed on slopes. Surface erosion of slopes can be controlled by protecting bare soil surfaces and by slowing or diverting surface runoff through the application of reinforced turf mats, geotextile and mulch blankets, hydro-seeded grass cover, and surface drains. Channelized surface erosion can be retarded with gully fills or plugs of cut brush or rock debris, or small check dams.

Intensive slope-stabilization and erosion-control techniques such as many of those listed above may be too costly for large areas of volcanically disturbed drainage basins, but they may be cost-effective in specific problem areas. Over large areas, economically feasible approaches may include tree planting, grass seeding, and grazing management to limit further destruction of slope-stabilizing vegetation. However, much post-disturbance erosion is likely to occur before grass seed can germinate or tree seedlings can grow to effective size, and a number of studies have shown that large-scale aerial grass seeding is no more effective for erosion control than the regrowth of natural vegetation (deWolfe et al. [2008]).

Lake stabilization or drainage

Stabilizing or draining lakes that could breach catastrophically without warning is another way to prevent lahars from reaching vulnerable downstream areas. Crater lakes, debris-dammed lakes (dammed by pyroclastic-flow, debris-avalanche, or lahar deposits), and glacial moraine-dammed lakes all can become unstable if their impounding natural dams are overtopped or structurally fail. Historic rapid lake outbreaks in several countries have triggered catastrophic lahars that resulted in loss of life (O'Shea [1954]; Neumann van Padang [1960]; Umbal and Rodolfo [1996]; Manville [2004]). Very large prehistoric outbreaks of a volcanically dammed lake have been documented having peak flows comparable to the world’s largest floods (Scott [1988]; Manville et al. [1999]). Stabilization methods include armoring of existing spillways on natural dams, construction of engineered spillways, and rerouting lake outflow by pumping or drainage through tunnels (Sager and Chambers [1986]; Willingham [2005]) (Figure 6; Case study 2). Preemptive drainage of dangerous lakes can be fraught with difficulties and may not be successful (Lagmay et al. [2007]).

Case study 2. Examples of lake stabilization

Since AD 1000, 27 eruptions of Mount Kelud (Java, Indonesia) have catastrophically expelled lake water from the volcano’s crater lake and created several deadly lahars, including a lahar in 1919 that killed more than 5000 people (Neumann van Padang[1960]). In an attempt to drain this lake, engineers in 1920 dug a drain tunnel over 955 m in length from the outer flank of the cone into the crater but eventually abandoned the project because of ongoing volcanic activity and other technical difficulties. Thereafter, siphons were constructed to control the lake level, and these were responsible for partial drainage of the crater lake and for a reduced number of lahars during the 1951 eruption (Neumann van Padang[1960]).

More recently, debris-avalanche and pyroclastic-flow deposits from the 1980 eruption of Mount St. Helens (Washington, USA) blocked tributary drainages of the North Fork Toutle River and enlarged several preexisting lakes. The largest and potentially most dangerous of these was Spirit Lake, which, when mitigation efforts began, was impounding 339 million m ³ of water—enough to form a lahar that could have destroyed major parts of several cities located approximately 90 km downstream. To prevent the Spirit Lake blockage from ever being breached by overflow, the level of the lake surface was stabilized by the U.S. Army Corps of Engineers (USACE) at a safe level, first by pumping water over the potentially unstable natural dam in pipes using diesel pumps mounted on barges, and thereafter by draining lake water through a 3.3-m-diameter outlet tunnel that was bored 2.5 km through an adjacent bedrock ridge to form a permanent gravity drain that was completed in 1985 (Figure6). The USACE stabilized the outlets from two other debris-dammed lakes at Mount St. Helens (Coldwater and Castle Lakes) by constructing engineered outlet channels. The Spirit Lake drainage tunnel continues to function well, although periodic inspection and maintenance of the tunnel are necessary. None of the stabilized lakes at Mount St. Helens have had outbreaks (Sager and Budai[1989]; Willingham[2005]).

Lahar diversion

Lahars can be prevented from spreading out and depositing in critical areas by keeping them channelized in modified natural channels or by engineering new channels. Such artificial channels (Figure 7a) must be sufficiently smooth, steep, and narrow (to maintain sufficient flow depth) in order to prevent in-channel deposition. The goal of such channelization is to keep lahars flowing so that they bypass critical areas. The effectiveness of this approach depends on lahar size and composition, channel dimensions, and construction techniques. Highly concentrated lahars (debris flows) can transport large boulders at high velocity and are extremely erosive, so channel bottoms and sides must be lined with concrete or stone masonry surfaces. Even so, hardened diversion channels may require frequent maintenance. Without hardening, lahars in diversion channels can easily erode channel boundaries and establish new flow paths. Channelization of lahar-prone streams draining volcanoes is relatively common in Japan and Indonesia (Smart [1981]; Japan Sabo Assoc. [1988]; Chanson [2004]).

Deflection and diversion structures also can be employed to reroute or redirect lahars away from critical infrastructure or communities. Structures include (a) tunnels or ramps to direct flows under or over roads, railroads, and pipelines; (b) training dikes (also termed levees or bunds) oriented sub-parallel to flow paths to guide lahars past critical areas; and (c) deflection berms oriented at sharper angles to flow paths to force a major course alteration in a lahar (Baldwin et al. [1987]; Hungr et al. [1987]; Huebl and Fiebiger [2005]; Willingham [2005]). However, lahar diversion may cause additional problems (and political resistance) if the diversion requires the sacrifice of only marginally less valuable land. Diversion ramps and tunnels are more practical for relatively small flows, whereas training dikes and deflection berms can be scaled to address a range of lahar magnitudes.

Dikes and berms are constructed typically of locally derived earthen material, but to be effective, these structures must be revetted (armored) on surfaces exposed to highly erosive lahars (Figure 7b). Revetment can be accomplished with thick layers of poured-in-place reinforced concrete, heavy concrete blocks or forms, heavy stone masonry faces or walls, stacked gabions, or steel sheet piles; layers of unreinforced concrete only centimeters thick cannot withstand erosion by large lahars (e.g., Paguican et al. [2009]). However, if a well-revetted dike is overtopped, rapid erosion of the unarmored back side of the dike can quickly cause dike failure and breaching nontheless (Paguican et al. [2009]) (Case study 3). In Japan, where probably more of these structures are constructed than anywhere else in the world, a major design criterion is that their orientation should ideally be less than 45° to the expected attack angle of a lahar to minimize overtopping and erosional damage (Ohsumi Works Office [1995]). Sometimes emergency levees are constructed without revetments, but this usually results in unsatisfactory performance, sometimes with disastrous results (Case study 1).

Case study 3. Lahar and sediment containment and exclusion structures

In the months following the May 18, 1980 eruption of Mount St. Helens (Washington, USA), the U.S. Army Corps of Engineers (USACE) built a rock-cored earthen sediment-retention structure (N-1 sediment dam) as a short-term emergency measure to try to hold back lahars and some of the volcanic sediment expected to wash downstream (Willingham[2005]). The structure had two spillways made of rock-filled gabions covered with concrete mortar; it was 1,860 m long and 13 m high, and was located approximately 28 km downstream of the volcano. Neither the upstream nor downstream face of the dam was revetted. Within a month of completion, one of the spillways was damaged by high flow. That spillway was repaired and resurfaced with roller-compacted concrete. In slightly more than a year, the N-1 debris basin filled with about 17 million m ³ of sediment, and the bed of the river aggraded nearly 10 meters. During the summer of 1981, the USACE excavated 7.4 million m ³ from the debris basin, but the river replaced that amount and added more during the following winter. The dam was overtopped and breached in quick succession by two events in early 1982—a major winter flood in February and an eruption-triggered, 10-million-m ³ lahar in March. Overtopping caused deep erosion of the downstream face of the dam at several points, which led to breaching. Even the reinforced, roller-compacted concrete spillways were scoured tens of centimeters, exposing ends of steel reinforcing bars that were abraded to dagger-like sharpness. The extensive damage to the dam and the limited capacity of the catch basin resulted in abandonment of the project (Pierson and Scott[1985]; Willingham[2005]).

Several years later, the USACE started construction of another larger sediment-containment dam (the Sediment Retention Structure or SRS), which was completed in 1989 and further modified in 2012 (Figure 8 a). It was built 9 km downstream of the original N-1 structure. In addition to trapping fluvial sediment, it was also designed to intercept and contain a possible future lahar (estimated peak discharge up to 6000 m ³ /s) from a potential breakout from Castle Lake. The SRS is a concrete-faced (upstream face), rock-cored, earthen dam about 550 m long, 56 m high, 21 m wide at the crest, and has a 122-m-wide armored spillway; its upstream catch basin is 13 km ² in area and was designed to hold back about 200 million m ³ of sediment (USACE—Portland District, unpublished data). By 2005, infilled sediment reached the level of the spillway, and river bed-load sediment began to pass through the spillway, even though the catch basin was filled only to 40% of estimated capacity. After 2005, only a fraction of the river’s sediment load was being intercepted, so raising of the spillway by an additional 2.1 m was completed in 2012 and experiments are continuing to induce greater sediment deposition in the upstream basin. The SRS has performed an important function in preventing large amounts of sediment from reaching and filling a reach of the Cowlitz River farther downstream and thus preventing serious seasonal flooding in communities along that river. No attempt has yet been made to excavate and remove sediment from behind the SRS.

An example of a lahar exclusion structure is the levee system enclosing the Drift River Oil Terminal (DROT) in Alaska (USA), which is a cluster of seven oil storage tanks that receive crude oil from Cook Inlet oil wells via a pipeline, plus some buildings and an air strip (Dorava and Meyer[1994]; Waythomas et al.[2013]). The DROT is located on the broad, low-gradient flood plain at the mouth of the Drift River, about 40 km downstream of Redoubt Volcano (Figure8c). Oil is pumped from these tanks to tankers anchored about 1.5 km offshore at a pumping-station platform. A U-shaped levee enclosure (built around the DROT but open at the downstream end) was raised to a height of 8 m following the 1989–1990 eruption, in order to increase protection of the facility from lahars and flooding. During both the 1989–1990 and 2009 eruptions of Redoubt, lahars were generated that flowed (at low velocity) up against the levees. Minor overtopping of the levees and backflow up from the open end caused some damage and periodic closure of the facility. The river bed aggraded to within 0.5 m of the levee crest in 2009, and the levees were thereafter reinforced and raised higher. The levee enclosure basically did its job, though it would have been more effective if the enclosure had been complete (on four sides).

Lahar containment or exclusion

Various structures can prevent lahars from reaching farther downstream, or seal off and protect critical areas while surrounding terrain is inundated. Sediment retention dams (Figure 8a) or containment dikes are used hold back as much sediment as possible but not necessarily water. To contain lahars, they must be constructed to withstand erosion and possible undercutting along their lateral margins and be tall enough to avoid overtopping. Under-design of these structures or inadequate removal of trapped sediment behind them can result in eventual overtopping and failure of the structure (e.g., Paguican et al. [2009]; Case study 3). The area upstream of a barrier where sediment is intended to accumulate is usually termed the catch basin or debris basin. Small excavated catch basins are also termed sand pockets. Such accumulation zones are typically designed to accommodate sediment from multiple flow events, and large tracts of land may be needed for this purpose. However, acquisition of land for this purpose can be problematic (Case study 1). If the design capacity is not large enough to accommodate all of the sediment expected to wash into a catch basin, provisions must be made to regularly excavate and remove accumulated sediment.

In addition to specially built lahar-related structures, pre-existing dams can sometimes be useful in containing all or most of the debris in a lahar (Figure 8b). Dams built for flood control or for impoundment of water for hydroelectric power generation or water supply can contain lahars and prevent them from reaching downstream areas, as long as (a) sufficient excess storage capacity exists behind the dam to accommodate the lahar volume, and (b) there is no danger of lahar-induced spillover at the dam in a way that could compromise dam integrity and lead to dam failure. Reservoir drawdown during volcanic activity might be necessary to ensure sufficient storage capacity to trap a lahar. This was done at Swift Reservoir on the south side of Mount St. Helens prior to the 1980 eruption, allowing it to successfully contain two lahars totaling about 14 million m³ (Pierson [1985]).

Exclusion dikes can enclose and protect valuable infrastructure, as was done in 1989–1990 and 2009 to protect oil storage tanks at the mouth of the Drift River, Alaska, from lahars and volcanic floods originating from Redoubt Volcano (Dorava and Meyer [1994]; Waythomas et al. [2013]) (Case study 3; Figure 8c). Diked enclosures may be a more appropriate strategy than channelization, diversion, or deflection in areas with low relief where low channel gradients encourage lahar deposition and where areas to be protected are small relative to the amount of channelization or diking that otherwise would be required.

Check dams to control lahar discharge and erosion

Some structures are built to slow down or weaken lahars as they flow down a channel. Check dams are low, ruggedly built dams that act as flow impediments in relatively steep stream channels (Figures 9 and 10). They have four functional roles: (a) to prevent or inhibit downcutting of the channel, which in turn inhibits erosion and entrainment of additional sediment; (b) to trap and retain some of a lahar’s sediment, thereby decreasing its volume; (c) to add drop structures to the channel profile in order to dissipate energy and slow downstream progress of the lahar; and (d) to induce deposition in lower-gradient reaches between dams (Smart [1981]; Baldwin et al. [1987]; Hungr et al. [1987]; Johnson and McCuen [1989]; Armanini and Larcher [2001]; Chanson [2004]; Huebl and Fiebiger [2005]; deWolfe et al. [2008]).

Check dams are commonly built in arrays of tens to hundreds of closely spaced dams that give a channel a stair-step longitudinal profile. Very low check dams are also called stepped weirs and are commonly constructed between larger check dams to act as hydraulic roughness elements for large flows (Chanson [2004]). A variety of styles and sizes of check dams have been developed, but fall into two basic categories: permeable or impermeable.

Permeable slit dams, debris racks, and open-grid dams (Figure 9a) are constructed of heavy tubular steel or structural steel beams, commonly with masonry bases and wing walls. Such structures are designed to act as coarse sieves, catching and retaining boulder-size sediment in a lahar but allowing finer material and water to pass through with depleted energy and mass. In addition to reducing the velocity of flows as they pass through, these dams also attenuate peak discharge. The effect is most pronounced on granular (clay-poor) debris-flow lahars that typically have steep, boulder-laden flow fronts. A variation on these vertically oriented structures is the drain-board screen (Azakami [1989]) (Figure 9b), which is a horizontally oriented steel grate or grill that performs the same sieving function for boulders as permeable dams when a lahar passes over the top of the grate, retaining coarse clasts while water and finer sediment drop down through the grate. Because of their orientation, these structures do not have to withstand the same high lateral forces as the upright permeable dams.

Impermeable check dams are composed of solid concrete, concrete with a packed earthen core, or steel cribs or gabion baskets filled with rocks and gravel (Figure 10). They may have small slits or pipes to allow exfiltration of water through the dam, in order to minimize impoundment of water. Gabions are used widely in the developing world because of their low construction costs—gravel fill often can be excavated locally from the channel bed, their permeability, and their flexibility, which can allow a dam to sag without complete failure if undermined by erosion. The crests of impermeable check dams commonly slope toward the center of the dam, where a notch or spillway is constructed, in order to direct streamflow or lahars over the dam onto a thick concrete apron extending downstream to protect the toe of the dam from erosion. Concrete sills or roughness elements commonly are placed at the downstream ends of aprons to further slow the flow that passes over the main dam. If upstream catch basins fill to capacity with sediment, check-dam functions are then limited to a, c, and d noted above, but full functionality can be restored if catch basins are regularly excavated.

‘Low-Tech’ warning systems

In some developing countries, effective low-tech warning systems employ human observers to alert threatened populations. Observers can be positioned at safe vantage points within view of lahar-prone river channels at times when flows have a high likelihood of occurring, such as during ongoing eruptions and during and following intense rainfall, particularly within the first few years after eruptions (de Bélizal et al. [2013]; Stone et al. [2014]). Observers stationed near lahar source areas are in a position to see or hear localized convection-cell rain storms that can trigger lahars, and human hearing can be very effective in detecting the approaching lahars themselves, often minutes before they come into view. The low-frequency rumbling sound caused by large boulders grinding against the river bed can carry hundreds or thousands of meters through the air and through the ground—a sound that is unmistakable to a trained observer. For example, a relatively small lahar occurring recently at Mount Shasta, California, sounded “like a freight train barreling down the canyon” and at times “like a thunder rumble” to a U.S. Forest Service climbing ranger (Barboza [2014]).

Once a lahar is detected, an observer can quickly issue an alert directly (by drum, siren, cellular phone, hand-held radio, etc.) to people living nearby (Figure 11a). This basic approach to lahar detection may be preferable where there is limited technical or financial capacity for maintaining sensors and other electronic equipment, where there are safe and accessible observation points, where there is high likelihood of expensive instruments being damaged or stolen without someone to guard them, where environmental conditions are challenging, or where electrical power and telecommunications are unreliable. Lahar detection by human observers is not immune to failure, however. Reliability is a function of the trustworthiness and alertness of the observers, their level of training, and the effectiveness of the alert notification method.

Automated telemetered warning systems

Automated electronic warning systems can be used to detect approaching lahars and telemeter alerts in areas where electrical power, technical support capabilities, and funding are more assured. Systems also can be designed to detect anomalous rainfall or rapid snowmelt that could trigger lahars, sense incipient motion of an unstable rock mass or lake-impounding natural dam, or detect an eruption that could trigger a lahar (Marcial et al. [1996]; Sherburn and Bryan [1999]; LaHusen [2005]; Manville and Cronin [2007]; Leonard et al. [2008]; USGS [2013]) (Figure 11b). In order for data from any of these various sensors to be useful for alert notification, they must be transmitted from remote sites in real time to a receiving station. Transmission can be accomplished by either ground-based or satellite-based radio telemetry (LaHusen [2005]) or cellular phone (Liu and Chen [2003]). Alert notifications can occur either automatically when some threshold in the level of the detection signal is exceeded, or an intermediate step can involve emergency management personnel, who verify and validate the detection signal before an alert is issued. Coordination among multiple agencies is critical to the success of an automated system, because hardware and software development of the sensor and the data acquisition/transmission systems are typically handled by physical scientists and engineers, whereas the development, operation, and maintenance of warning systems are typically managed by emergency managers and law-enforcement personnel (Case study 4).

Case study 4. The Mount Rainier lahar warning system

A significant volume of rock on the upper west flank of Mount Rainier (USA) has been extensively weakened (60–80% loss in unconfined strength) by hydrothermal alteration and is unstable (Watters et al.[2000]; Finn et al.[2001]; John et al.[2008]). A lahar warning system was developed by the U.S. Geological Survey and Pierce County (Washington) to detect potential lahar initiation from this sector, and it was installed in 1995 by USGS and Pierce County personnel in the Carbon and Puyallup River valleys downstream of the weak and oversteepened rock mass (USGS[2013]). The system is designed to warn tens of thousands of people who live in the downstream lahar hazard zone of an approaching lahar. Affected communities are situated from 40 to 80 km downstream of the volcano and could have from 12 minutes to 2 hours, depending on location, to evacuate after receiving a warning message. Since installation, the warning system has been maintained and operated by the Pierce County Department of Emergency Management, in collaboration with the Washington State Emergency Management Division.

The system comprises specialized seismic sensors capable of detecting ground vibrations within a frequency range typical of lahars (30–80 Hz), a ground-based radio telemetry system for detection-signal transmission, and a combination of sirens, direct notification, and the Emergency Alert System (EAS) that utilizes NOAA weather radios for warning message dissemination (LaHusen[2005]; USGS[2013]). County and state emergency-management agencies and city and county law-enforcement agencies collectively have responsibility for verifying and validating alerts from the sensors, activating warning sirens, and sending warning messages.

Collaboration between all the agencies involved in lahar hazard warning and risk reduction at Mount Rainier is fostered by regular meetings of the “Mount Rainier Work Group”. Such lahar warning systems require ongoing collaboration between scientists and emergency management officials, as well as regular maintenance and testing. Members of the at-risk population (including schools) have been assigned evacuation routes, have been informed about what to do when a warning message is received, and regularly participate in evacuation drills (Figure 3g).

Warning message development and delivery

In the simplest warning systems, warning messages are delivered only as simple audible signals (drums, sirens, whistles, etc.), and the affected population must be informed beforehand about what the signals mean and what the appropriate response should be. In more sophisticated systems, incident-specific alert messages can be delivered to large populations simultaneously by cellular phone, the Internet, radio, or television. In these cases, the alert must convey a definitive and unambiguous message that effectively prompts individuals to take protective actions. Several factors influence the effectiveness of a warning message, including the content and style of the message, the type and number of dissemination channels, the number and pattern of warning statements, and the credibility of the warning source (Mileti and Sorenson [1990]).

Warning messages should be specific, consistent, certain, clear, and accurate (Mileti and Sorenson [1990]). To ensure credibility, message content should include a description of the hazard and how it poses a threat to people, guidance on what to do to maximize personal safety in the face of impending danger, location of the hazard, the amount of time people have to take action, and the source of the warning. The more specific a warning message is, the more likely the receiver is to accept the warning (Cola [1996]; Greene et al. [1981]). Emergency warnings without sufficient detail create information voids, and the affected population may then rely on ill-informed media commentators, friends, neighbors, or personal bias and perceptions to fill this void (Mileti and Sorenson [1990]). Input from volcano scientists is critical for some of this detail and specificity.

Both credibility and consistency of the warning message are important. At-risk populations commonly receive information from informal sources (for example, the media, friends, social media), sometimes more quickly than through various official channels during a crisis (Mileti [1999]; Leonard et al. [2008]; Dillman et al. [1982]; Mileti and Sorenson [1990]; Parker and Handmer [1998]; Mei et al. [2013]). For example, 40–60% of people in the vicinity of Mount St. Helens first received informal notification of the 1980 eruption (Perry and Greene [1983]; Perry [1985]). The proliferation of informal information channels today with the Internet and social media can benefit the warning dissemination process, because individuals are more likely to respond to a warning if it is confirmed by multiple sources (Cola [1996]; Mileti and Sorenson [1990]). But multiple sources become problematic if they advance conflicting information, causing individuals to become confused. Therefore, challenges for emergency managers and scientists are to keep reliable information flowing quickly and to maintain consistent messages, both during and after an emergency. Joint information centers can ensure that (a) there is consistency in official warning statements among multiple scientific and emergency-management agencies, (b) easy access is provided for the media to the official information and to experts who can explain it, and (c) the effectiveness of warning messages is monitored (Mileti and Sorenson [1990]; Driedger et al. [2008]).

Evacuation training

Warnings are given so that people in a lahar flow path can move quickly out of harm’s way. Sheltering in place is generally not a viable option. The lives of at-risk individuals may depend on understanding that they are living in, working in, driving through, or visiting a lahar hazard zone, as well as understanding what to do when they receive a warning (Mileti and Sorenson [1990]; Leonard, et al. [2008]). As the world witnessed in the 1985 Nevado del Ruiz disaster (Voight [1990]) (Case study 5), warnings that a lahar was bearing down on their town were not able to prevent catastrophic loss of life, because the warnings were issued without the population’s understanding of the risk or how they should respond. To increase the likelihood of successful evacuations, scientists should encourage and help lead hazard-response exercises and evacuation drills, especially in areas with short time windows for evacuating hazard zones. These exercises and drills provide emergency managers the opportunity to identify weaknesses in the warning–evacuation process and to minimize potential delays that could result from confusion, insufficient information, or lack of understanding on what to do. They also provide scientists with a platform for discussing past catastrophes and the potential for future events. Holding an annual table-top exercise or community-wide evacuation drill on the anniversary of a past disaster can help to institutionalize and personalize the memory of past events, an important step if new community members are to take these threats seriously. A well-educated and trained community that possesses information about where they will get information and what emergency actions to take is less likely to be confused by warning messages, to resist evacuation orders, or to blame officials for ordering an evacuation when a catastrophic event fails to occur (e.g., Cardona [1997]). The goal for scientists and emergency managers is to create a “culture of safety” (cf., Wisner et al. [2004], p. 372) where at-risk individuals understand potential hazards, take personal responsibility for reducing their risks, understand how to respond to an event, and realize that lessening of risks requires actions from all levels of a community and government.

Case study 5. The Nevado del Ruiz disaster

The 1985 Nevado del Ruiz lahar disaster, which cost approximately 21,000 lives in the town of Armero, Colombia (Figure1a), is an excellent case study of the complexities that can lead to ineffective evacuation after warning messages are broadcast, poor emergency response, and a haphazard disaster recovery (Voight[1990]; Hall[1992]). In post-event analyses, it was generally concluded that the Ruiz catastrophe was the result of cumulative human and bureaucratic errors, including lack of knowledge, misunderstanding and misjudgment of the hazard, indecision, and even political barriers to effective communication, rather than inadequate science or technical difficulties. Other factors contributing to the catastrophe included evacuation plans that had been prepared but not shared with the public, poorly equipped emergency management authorities, the absence of agreed-upon decision-making processes, and uncertainty about the pre-event hazard assessments that made public officials reluctant to issue an early evacuation order because of the potential economic and political costs. The hazard maps produced by scientists for Nevado del Ruiz prior to the eruption were highly accurate in their predictions of where lahars could go, but they were published only about a month before the disaster, giving little time for assimilation and responsive action by the emergency managers. Furthermore, production of the maps did not lead to effective risk communication, because the scientists who made the maps generally did not engage in conveying that risk information in understandable terms to officials and the public. Scientists may prepare excellent hazard assessments and maps, but unless they participate fully in conveying hazard information to officials and the public in ways that are understandable, disasters can still happen (Voight[1990]; Hall[1992]).