1 Introduction

Reduced visibility may occur during primary volcanic ashfall or through the remobilisation and resuspension of existing fall deposits (Sparks et al. 1997; Baxter and Horwell 1999; USGS 2013; Folch et al. 2014) into the atmosphere by wind, vehicle movement, cleaning processes or other human activities. Both primary and remobilised ash may produce potential issues for transportation (Barsotti et al. 2010; Folch 2012; Wilson et al. 2014; Blake et al. 2016a, b, 2017a, b). There have been many cases where reduced visibility following volcanic eruptions has impacted surface transportationFootnote 1 function, particularly road (Table 1), but also rail, maritime and at airports, sometimes to near-total darkness (Blong 1982; Guffanti et al. 2009). For example, following the Mount St Helens eruption in 1980, “zero visibility” caused traffic to come to a standstill in several places including on Interstate 90 near Ellensburg in Washington, with visibility so poor that the authorities used flares to guide motorists to nearby schools and churches (Warrick 1981, p.19). Once the initial ashfall was over, it became imperative to control traffic movement through speed restrictions, spacing vehicles on roads and road closures to avoid vehicles creating “great clouds of ash” and again reducing visibility on the roads (Warrick 1981, p.20). Hundreds of accidents in the affected areas were attributed to the billowing of ash behind vehicles and associated loss in visibility (Blong 1984). More recently, following the eruption of Cordón Caulle, Chile, in 2011, visibility in neighbouring Argentina, where ash fell, was so low that several main roads including Route 40 (linking Bariloche city and the Neuquén province) were subsequently closed. This affected the transportation of people and freight at a national level (Folch et al. 2014). Vehicle headlights and brake lights are often reported to be ineffective in such conditions and barely visible to other drivers (Cole and Blumenthal 2004; Wilson et al. 2012a; USGS 2013). Disruption to rail and maritime transportation caused by poor visibility during ashfall and resulting in safety and navigational issues has also been documented (Johnston 1997; Wilson et al. 2012a). For example, rail operations were shutdown during the initial Mount St Helens 1980 ashfall due to reduced visibility (as well as health concerns), with substantial speed restrictions and inspections necessary in the following days (Warrick 1981; Blong 1984).

Table 1 Historical records of visibility impacts on road following volcanic eruptions. Other occurrences described as general impacts to transportation, where impacts on visibility may have occurred but are not specified in the literature, are not shown
Full size table

Despite the many impacts to surface transportation that have been documented (Table 1), there has been little quantitative analysis into the effects of airborne volcanic ash on reducing visibility and the subsequent consequences for network functionality and safety. Contemporary critical infrastructure management increasingly aims to optimise network performance during natural hazards, especially for non-damaging events, such as minor ashfalls (Wilson et al. 2014). Ash thickness has usually been the key measure of hazard intensity adopted in the past (Table 1), but is likely to be of little relevance when assessing visibility impacts. Airborne ash concentration and ash-settling rates are more important for such impacts. The lack of an evidence base implies that surface transportation operators typically can only make crude management decisions of either taking a precautionary approach of shutting systems down in the presence of ash (or in the event of forecasted ash), or a reactive approach if airborne ash causes problems for a surface transportation system. However, several contemporary ash dispersion and fallout forecasting models, which are outlined in detail by Scollo et al. (2008) and Folch (2012), can provide outputs that include atmospheric concentrations and settling rates (e.g. FLEXPART—Stohl et al. 1998, 2005; VOL-CALPUFF—Barsotti et al. 2008; FALL3D—Costa et al. 2006; Folch et al. 2009). Therefore, the capacity now exists for surface transportation vulnerability assessments to adopt these metrics. Furthermore, settling rate and airborne concentration are important metrics to consider for other impacts to transportation. For example, it has been determined that concentrations as low as 1 × 10−4 g m−3 can cause substantial damage to aircraft (Witham et al. 2007; Folch and Sulpizio 2010).

In this paper, we next present a background section which summarises the interdisciplinary literature including the physical properties of volcanic ash, visibility degradation and surface transportation impacts. The background section also includes a summary of our Auckland case study site, which gives context to the work through the application of experimental processes and replication of ash types/colours, particle sizes and settling rates expected in the city from various volcanic sources. We then outline the methods used to simulate volcanic ashfall in a laboratory setting to investigate the effect of ash characteristics and settling rate on visibility degradation. Precise and consistent ash-generation rates (g h−1) are produced in a purpose-built container using a Solid Aerosol Generator (Topas SAG 410). We also use a dual-pass transmissometer (Dynoptic DSL-460 MkII) for the measurement of opacity as a proxy for particulate emissions. We use datasets containing ash characteristic and geospatial information from the literature available following worldwide historical eruptions to test our methodology, specifically focusing on the Auckland case study. Results from the experimental tests are presented before a discussion of the findings related to visibility in airborne ash and forecasts for Auckland. Finally, we discuss key limitations of the experimental approach, provide suggestions for future work and suggest evidence-based, semi-quantitative transportation mitigation strategies that could be implemented in operational environments.

2 Background

2.1 Ash deposit thickness and visibility relationship

Ash deposit thickness is the most frequently (and often only) reported hazard intensity metric following volcanic eruptions (Pyle 1989; Wilson et al. 2012a, 2014, 2017; Blake et al. 2017a, b). There is generally a decreasing trend in ash thickness with distance from eruptive vents, although other spatial variations are possible due to ash aggregation and/or an unusually large proportion of mid-sized particles, which can lead to secondary maxima on a thickness versus distance plot (see Carey and Sigurdsson 1982; Pyle 1989; Sparks et al. 1992; Scasso et al. 1994; Bonadonna et al. 1998; Bonadonna and Phillips 2003; Bonadonna and Houghton 2005; Costa et al. 2009; Parfitt and Wilson 2009; Bonasia et al. 2012; Folch et al. 2010; Macedonio et al. 2016). Ash deposit thickness is clearly an important parameter for certain transportation impact types at quite low levels. For example, line marking coverage and loss of skid resistance on roads can occur with 0.1–1.0-mm-thick ash in some situations (Blake et al. 2016a, b, 2017a, b). However, ash thickness is generally a poor indicator for estimating likely effects on visibility. Some studies indicate that thickness plays a minimal role in controlling airborne ash concentration, initial levels of visibility degradation and associated impact to transportation (e.g. Warrick 1981; Blong 1982). For example, reporting on the Mount St Helens 1980 eruption, Warrick (1981) states that “ash depth had little influence on the initial level of impact on the transportation systems”. Conversely, other studies suggest that thickness can be used to inform visibility impacts to a degree (e.g. Thorarinsson 1971; Johnston 1997; Searl et al. 2002). While there is potential for uncompacted fine ash to become airborne by disturbance of the deposit, and thicker deposits can prolong the duration of such remobilisation and subsequent recovery time (Searl et al. 2002; Hincks et al. 2006; Wilson et al. 2011), we propose that ash-settling rateFootnote 2 (especially for direct ashfall) and airborne particle concentrationFootnote 3 (for both direct and resuspended ash) are more appropriate metrics to adopt when assessing visibility impairment in environments containing volcanic ash. Therefore, the thicknesses outlined in Table 1 have little direct link to the visibility observations and should only be used as a loose proxy for airborne ash concentration and duration of exposure.

2.2 Volcanic ash-settling rate

Two fundamental processes affect ash-settling rate, which are depicted in Fig. 1:

Fig. 1
figure 1

Eruption and environmental characteristics that influence volcanic ash-settling rates and airborne concentrations (eruption column characteristics modified from Sparks 1986; Carey and Bursik 2015)

Full size image
  1. 1)

    Primary ashfall from the eruptive vent, which is the result of explosive volcanic eruptions causing the disintegration of magma or vent material and production of rock fragments (Jenkins et al. 2014). The terminal fall velocityFootnote 4 of sedimenting particles affects ash-settling rate and is mainly dependent on their diameter, densities and, to some extent, their shape (see: Bonadonna et al. 1998; Riley et al. 2003; Parfitt and Wilson 2009; Pardini et al. 2016).

  2. 2)

    The remobilisation and resuspension of existing ash into the atmosphere from entrainment by meteorological winds, small-scale atmospheric turbulence caused by vehicle movement, cleaning processes and/or other human activities. The quantity of volcanic ash which becomes resuspended is dependent on the particle size, density, water content, shape, roughness of the ground material on which it has settled, and the degree of binding and compaction (Fowler and Lopushinsky 1986; Sivakumar 2005; Wilson et al. 2011)

We suggest that different ash-settling rates can be established under laboratory conditions. However, compared to other hazard intensity metrics such as ash thickness and particle size, there are somewhat limited primary empirical data available for volcanic ash-settling rates obtained from field measurements or indeed from records of ashfall duration associated with ash thickness or loading in the field. Mean settling rates can be estimated based on the total ashfall duration (e.g. see: Hill 2014), although it is possible that pulses of ashfall with periods of quiescence occur during that time (Scasso et al. 1994). Furthermore, complications from plume and atmospheric turbulence, particle–particle interaction and atmospheric conditions (Bonadonna et al. 2011) mean that accurate correlations between total ash thicknesses and settling rates are difficult to produce.

2.3 Particle size

Unlike airborne mineral dust, which often has a homogeneous internal structure consisting of similar-sized particles, volcanic ash is generally inhomogeneous; the size of erupted material may vary by several orders of magnitude (Weinzierl et al. 2012). However, the most important volcanic ash particle size range in terms of impact to infrastructure including transportation is generally fine-grained ash (e.g. Wilson et al. 2012a; Blake et al. 2016b) (i.e. often < 100 μm diameter, and particularly < 64 μm diameter; Carey and Sigurdsson 1982). Fine ash can disperse for up to hundreds of kilometres from the vent before settling due to gravity and therefore generally settles over extensive areas. The proportions of ash in these fine fractions increase with increasing eruption explosivity (Moen 1981; Horwell 2007; Durant et al. 2009; White et al. 2011), and fine ash is more susceptible to remobilisation by the processes described in point 2 above (Wilson et al. 2011). Particle size distributions of volcanic deposits are poorly constrained due to sparse data, variations in wind conditions and dimensions of eruption columns, and inconsistencies in the methods of measurement. Many particle size analyses carried out for ash deposits are incomplete, lacking data below 63 µm (Bonadonna and Houghton 2005).

2.4 Particle shape

The less spherical a particle shape of the same weight, the lower its settling rate (Komar and Reimers 1978; Wilson and Huang 1979). Volcanic ash particles are often more elongated in shape than other particulates such as mineral dust. Wilson and Huang (1979) found that for volcanic particles of 30–500 µm diameter, glass and feldspar fragments have a very high proportion of flattened particles, whereas pumice consists of a greater variety of shapes (Fagents et al. 2013). More recently, Weinzierl et al. (2012) compared the physical properties of volcanic ash and mineral dust by using samples from Eyjafjallajökull Volcano, Iceland, and the Sahara Desert. The median aspect ratio (i.e. ratio of the width to the height) of Eyjafjallajökull ash was 2.0, compared to a median aspect ratio of 1.6 for the mineral dust. Therefore, because of the general absence of sphericity, settling rates of volcanic ash may be slower than expected on the basis of particle size and weight.

2.5 Measurement of visibility

Visibility involves human perception of the environment, and thus, no instrument truly measures visibility (Malm 1979). Aerosols in the atmosphere, including volcanic ash, interact with light waves leading to absorption and scattering. The amount of light energy redirected from its original path is referred to as the extinction coefficient (bext) and is equal to the sum of four interactions (Robinson 1968; Hyslop 2009):

$$b_{\text{ext}} = \, b_{\text{scat,p}} + \, b_{\text{scat,g}} + \, b_{\text{abs,p}} + \, b_{\text{abs,g}}$$
(1)

where bscat,p is the light scattering by particles, bscat,g is that by gases, babs,p is the light absorption by particles, and babs,g is that by gases.

The extinction coefficient is thus the optical parameter which is the best proxy for visibility assessment. Once the extinction coefficient is calculated, the corresponding visual range (VR) in metres can be estimated, defined as the longest distance that a large, black object can be seen against the sky at the horizon with the unaided eye (Binkowski et al. 2002; Seinfeld and Pandis 2006; Hyslop 2009; Blake et al. 2016a):

$${\text{VR }} = \, 3.912/\left( {b_{\text{ext}} } \right)$$
(2)

where the value at the numerator is constant.

Further information on the extinction coefficient and visual range can be found in Online Resource 1.

Particles smaller than the wavelength of visible light (particle diameter < 0.05 μm in white light) have little effect on visual range. Larger particles (particle diameter > 2 μm in white light), however, have a much greater effect with characteristics such as size, shape and composition being much more influential (Conner 1974; EPA 2000). Although some studies assess the optical properties of volcanic ash at high altitude using remote sensing techniques for the purposes of flying aircraft operability (e.g. Weinzierl et al. 2012), to the best of our knowledge, there have been no detailed quantitative studies of visual range during ashfall near ground level. However, if visibility characteristics associated with volcanic ash can be quantified near the ground, then comparisons with other, more studied atmospheric hazards such as fog and dust storms can be made, and impacts on surface transportation and vehicle mobility estimated. We suggest that where visibility in volcanic ash has not been quantified, precise comparisons to visibility during other atmospheric hazards cannot be made as the characteristics of certain particles and subsequent effect on light attenuation may be different.

2.6 Effects of visibility degradation on surface transportation networks

Through the common effect of visibility degradation, many atmospheric hazards produce similar consequences for surface transportation. Specifically, previous studies document that the presence of thick fog, smoke or dust influences roads by: (1) reducing the volume of traffic; and (2) increasing the risk of accidents despite lower traffic volumes and an overall reduction in speed—particularly accidents that involve multiple vehicles and cause a higher percentage of severe injuries (OECD 1986; Musk 1991; Ashley et al. 2015; Hardy 2015; Ibrahim 2015).

There have been many studies into the impacts of fog and mineral dust on visibility (e.g. Codling 1971; Moore and Cooper 1972; Hagen and Skidmore 1977; Summer et al. 1977; Perry 1981; Musk 1991; Taylor and Moogan 2010; Abdel-Aty et al. 2011; Weinzierl et al. 2012; USDOT 2013; Ashley et al. 2015). High concentrations of fog droplets and/or mineral dust particulates are required to cause low visibilities and disrupt transportation networks. For example, Hagen and Skidmore (1977) highlight that mineral dust concentrations exceeding 50 to 100 mg m−3 seriously reduce visibility during daylight, with lower concentrations hazardous at night. Other atmospheric hazards including smoke from wildfires and airborne volcanic ash, routinely disrupt road networks and inhibit operations at airports, and accidents have been attributed to volcanic ashfall (e.g. Bartney 1980; Folch et al. 2014; Blake et al. 2015). However, they remain relatively under-researched hazards in transportation studies (Cova and Conger 2003, Abdel-Aty et al. 2011).

A recent study by Brooks et al. (2011) used a driver-simulation method to determine how drivers react when driving in varying levels of fog and to assess whether drivers are willing to drive at speeds where their lane-keeping performance is degraded due to the reduced visibility. Observed speed reductions were as follows:

  • For visual ranges of 496–31 m, average speeds decreased from 91.3 to 89.1 km h−1

  • For 18 m visual range, speeds decreased further to 82.9 km h−1

  • In the foggiest condition tested (6 m visual range) speeds decreased to 71.7 km h−1, representing a decrease of ~ 20 km h−1 from clear conditions.

Despite the speed reductions for the most reduced visibility however, it was calculated that drivers would likely be incapable of stopping to avoid obstacles in the roadway (Brooks et al. 2011; Blake et al. 2016a), a situation that corresponds to what has been recorded on actual roads in inclement weather (Edwards 1999). Additionally, lane-keeping ability was reduced when fog resulted in visibility distances < 30 m (Brooks et al. 2011). Mueller and Trick (2012) also used a driver-simulation method to investigate average speeds in fog (with 600 m visual range) and found that all drivers reduced their speed in fog compared to clear conditions, with an average speed reduction of 6.4 km h−1.

Mitigation measures available for managing reduced visibility due to airborne volcanic ash include reducing vehicle speeds (including the implementation of lower-than-usual speed limits); restricting the number and/or type of vehicles on the network; increasing the space between moving vehicles; dampening surfaces with water to minimise the resuspension of ash, and closing roads in especially contaminated areas (Table 1). The only complete solution is the total removal of ash deposits, but this may not be possible or cost-effective.

2.7 Auckland case study

We use Auckland as a case study location to apply our methodology due to the relatively large population (~ 1.6 million people, Statistics New Zealand 2015), extensive transport networks and exposure to multiple sources and types of volcanic ash (Fig. 2).

Fig. 2
figure 2

Potential sources of volcanic ash in New Zealand that could affect Auckland

Full size image

Basaltic ash may originate from new eruptions in the Auckland Volcanic Field (AVF), on which the city is built, or potentially from Whangarei and Bay of Island volcanic fields to the north (Smith and Allen 1993). Andesite can be deposited from eruptions within the Tongariro Volcanic Centre, located in the Taupo Volcanic Zone (TVZ) and from Taranaki Volcano. Rhyolite may be deposited in Auckland from eruptions within the Taupo and Okataina Volcanic Centres of the TVZ with rhyolitic ash from Mayor Island also possible. Recent work by Zawalna-Geer (2016) suggests that Auckland has been inundated by tephra at least once every 424 years on average. However, any recurrence rates must be considered with caution due to different climatic regimes and because not all eruptions will be preserved in the geologic ash record, and certainly not all eruptions will be represented by ash at specific coring sites (Lindsay and Peace 2005). Auckland is exposed to ashfall with a relatively wide range of airborne concentrations, settling rates, as well as particle sizes, densities and shapes (Smith and Allen 1993). As such, many of the methods and results in this paper are internationally applicable.

Visual range for excellent conditions is typically taken to be > 70 km (MFE 2001). Indeed, the mean visual range taken over a three year period (from 2001 to 2003) from readings over Auckland City (Dr Gerda Kuschel, Auckland Council, pers comm, 21 June 2016) is similar, calculated at 66.5 km with 95% CI [65.1, 67.8]; the shorter visual range is unsurprising given the city environment with likely higher background particulate concentrations than “typical excellent” conditions.

3 Methodology

Our methodology outlines the novel technique used to simulate ashfall and to simultaneously measure ash-settling rates and extinction coefficients (subsequently used to calculate visual ranges using Eq. 2) in a laboratory environment. Visual ranges through ashfalls of different particle sizes and types and settling rates, representing those considered possible in Auckland, may thus be determined.

3.1 Experimental set-up

The key equipment used for our experimental set-up is a transmissometer to calculate visual range, and solid aerosol generator to disperse ash, both selected specifically for the purpose and incorporated into a custom-made experimental set-up.

3.1.1 Transmissometer

We use a dual-pass transmissometer (DynOptic DSL-460 MkII) for our experiments. The transmissometer is adopted to measure the optical characteristics over a preset path length of 1.4 m. It measures a specific amount of green light transmitted from the source/transceiver securely mounted to the side of a container, to a reflector mounted opposite and back to the same transceiver (Fig. 3). This allows the direct and continuous (1 s interval) calculation of the extinction coefficient of the air along the path length for light with a wavelength at which the human eye observes. A small amount of ash accumulates on the optics of the transmissometer during each test. However, following trial tests, this was substantially limited by extracting purge air (which the instrument uses to clean the optics during operation) through ducting from an adjacent clean room. Following the extraction of purge air from a clean source, the accumulation of ash on the optics accounted for only ~ 1.0 × 10−6% of the maximum values recorded which is deemed insufficient to substantially affect findings. The transceiver and reflector are simply cleaned with compressed air between each experiment and results corrected so as to always start with zero values for the extinction coefficient and avoid drift. We refer the reader to Online Resource 2 for further information on transmissometers, and to the DynOptic DSL-460 MkII operation manual (DynOptic 2014) for detailed information on the calibration and operation procedures for this particular transmissometer.

Fig. 3
figure 3

Experimental set-up developed and implemented for visibility testing (adapted from Blake et al. 2016a)

Full size image

3.1.2 Ash disperser

Precise and consistent ash-generation rates (g h−1) are produced using a solid aerosol generator (Topas SAG 410) (named ash disperser from herein) into the top and centre of a purpose-built stainless-steel cylindrical container (Fig. 3), shaped as such to encourage evenly distributed flow because trial tests suggested that cylinders were preferential to cuboids where ash-settling rate varied in corners.

Two simultaneous methods for calibration of the ash disperser for each ash type of particular particle diameter size distribution were used to determine the machine settings required for different generation rates and estimated ash-settling rates in the container; a gain-in-weight technique where ash dispersed from the instrument is collected in filter bags and weighed, and a loss-in-weight technique where the whole instrument weight is recorded before and after ash dispersal.

3.1.3 Ash-settling rate measurement

When the ash enters the container, a purpose-built nozzle on the end of the tube causes the ash to be dispersed in all directions before settling under gravity. The height of the container (1.3 m total, with a 1 m fall distance to the light beam of the transmissometer) was chosen by considering formulas derived from Bonadonna et al. (1998), which suggest that the majority of ash particles dispersed reach terminal velocity before passing through the measurement level. However, the set-up was also constrained by workspace limitations and accessibility requirements to the hopper of the ash disperser, which had to be positioned above. Beneath the measurement level of the transmissometer, the ash settles at the base of the container (0.75 m2 in area). Half of the area (i.e. 0.375 m2) at the centre of the base is dedicated to a circular board on top of an automated mass balance set to poll and record the weight of ash every 10 s. The other half around the edge consists of foam, which allows the air to evenly escape from the base of the container and the vast majority of ash to remain static when out of suspension. Trial tests suggested that the solid board at the centre had little influence on ash flow currents towards the base of the container and that there was minimal remobilisation of ash that would affect results.

Some radial differences in ash-settling rates were identified for samples, both for individual experiments and between experiments. However, the consistent airflow within the container, along with the light beam traversing the entire diameter, means that there is negligible effect on the transmissometer readings and that the entire horizontal transect would appropriately represent a steady ashfall. To account for the radial differences in settled ash at the base (i.e. because the board does not cover the whole area), a series of ten petri dishes are equally positioned (five on the central board and five on the foam edge). These are weighed after each experiment to analyse the distribution profile for the settling ash in our experiments. The results from the petri dishes, in conjunction with the centred mass balance weights, are used to calculate and record mean “actual” (rather than estimated) ash-settling rates (g m−2 h−1) at the base (Sect. 4.1).

Preliminary test runs revealed that equilibriums between ash generation into the container and ash settling at the base occurred within 15 min for all ash types. Therefore, every experiment for each ash type at a specific generation rate and distribution for particle diameter size is run for a total of 1 h where possible (some high generation rates caused ash supply to be exhausted within 45 min). This allows at least 30 min of continuous measurement at equilibrium conditions, which is sufficient time for a mean value of the extinction coefficient and visual range for each sample and ash-generation rate to be calculated. The transmissometer is then left operating for at least a further 15 min after the disperser is switched off to allow continued monitoring, whilst all the ash settles out. Ash is cleaned from the board, foam and container edges between each experiment and a thorough clean of all equipment is conducted. The ash disperser tube is disposed of and replaced at the end of testing with each ash sample.

3.2 Application to Auckland

Methods used to replicate ash types, particle sizes and settling rates that can be expected in the city given a future volcanic eruption in the AVF or from the larger volcanoes of the central North Island of New Zealand are described in Sects. 3.2.1–3.2.3, respectively.

3.2.1 Ash type

Basaltic, andesitic and rhyolitic ashfall is possible in Auckland, and we conduct experiments with all three types; a dark-coloured basalt which is sourced from a deposit of Pupuke Volcano in the AVF, a mid-coloured andesite derived from a deposit from the Poutu eruption of Tongariro Volcano, and a light-coloured rhyolite from the Kaharoa eruption of Tarawera Volcano, in New Zealand (Table 2). The raw samples were largely selected from sources outside of the AVF, as many of those within the AVF have been weathered, contaminated by organic material and/or disturbed or removed by human activity (e.g. Alloway et al. 2004; Cassidy and Locke 2004; Howe et al. 2011; Adams 2013). Our samples were modified through pulverisation to achieve a range of particle diameter size distributions possible in Auckland and thus likely contain a higher than natural proportion of particles that are blocky in nature with a high degree of angularity due to the milling process (Broom 2010). Therefore, an additional fine-grained sample (sourced from the 2008 eruption of Chaitén in Chile—Table 2) is also tested to investigate the effects of particle shape on visual range. This sample was not altered through pulverisation and only sieved to remove the larger particles that can cause clogging of the ash disperser. All samples were dried at 65 °C for > 48 h prior to testing.

Table 2 Characteristics of the four ash types used for experimentation
Full size table

3.2.2 Particle size

Following a detailed literature review, we further developed work by Hill (2014) to estimate ash particle sizes that can be expected in Auckland from volcanic sources in New Zealand (Fig. 4). Unlike Hill (2014) however, who focussed solely on median data, we incorporate all available data for particle size, including individual data or median size recorded at specific locations from a vent (shown as points in Fig. 4), modes (bold points or bold ranges), full distributions (vertical “error bars”) and distributions in spatial extent (horizontal “error bars”). Triangles in Fig. 4 denote maximum clast sizes of blocks, and black data points show particle sizes and distances from vents determined from recent analysis of deposits in six Auckland maars from historical AVF eruptions of One Tree Hill, Three Kings and Mt Eden Volcanoes (Jenni Hopkins, Victoria University of Wellington, pers comm, 19 July 2014).

Fig. 4
figure 4

Data from: 1Andò et al. (2014), 2Wilson et al. (2013), 3Heather Craig, University of Canterbury, pers comm, 17 July 2014, 4Watt et al. (2009), 5Bonadonna et al. (2011), 6Scollo et al. (2007), 7Bonadonna et al. (2002), 8Scasso et al. (1994), 9Harris and Rose (1983), 10Carey and Sigurdsson (1986), 11Scheidegger and Federman (1982), 12Sarna-Wojcicki et al. (1981), 13Carey and Sigurdsson (1982), 14Eto (2001), 15Wilcox (1959), 16Walker (1981), 17Sigurdsson et al. (1985), 18Jenni Hopkins, Victoria University of Wellington, pers comm, 19 July 2014

Particle sizes of ash deposits and their associated distances from 15 worldwide eruptive vents and analysis from maars in the AVF. The distances from central Auckland (depicted as 0 km) to New Zealand volcanic ash sources are identified by a series of dashed vertical lines, and approximate maximum axial extent of the AVF by a bold dashed vertical line. Solid horizontal grey lines denote the boundaries between ash and lapilli (2 mm) and lapilli and block (64 mm) particle diameters.

Full size image

Due to the maximum particle size constraints of the ash disperser and likely clogging when particle diameters exceed ~ 300 μm, our experiments focus on samples defined as fine ash by Carey and Sigurdsson (1982) and Folch et al. (2009), in that mode particle diameters for most samples are < 64 μm. All mode particle diameters are < 110 μm (Fig. 5). However, we subdivide our samples into three categories of mode particle diameter size for ease of interpretation; a. < 20 μm, b. 20–50 μm and c. > 50 μm (Table 2). For the basalt sample, we test ash for all three categories to investigate the effect of particle diameter size on visual range. For the andesite and rhyolite samples however, we only test for the category b range due to ash sample availability. Specific ash particle sizes were achieved using a rock pulveriser with different disc separation distance and sieves with different mesh aperture sizes, followed by laser sizing to determine the particle diameter size distributions. Total particle diameters for all samples used are ~ 1–320 μm (Fig. 5) which corresponds with the mid- to lower-grain sizes of ash particles that can be expected in Auckland given a future eruption in New Zealand (Fig. 4).

Fig. 5
figure 5

Particle diameter size distribution plots for the ash used in experimentation, derived from three tests per sample using a Micromeritics Saturn DigiSizer II laser sizer instrument. a–c refer to the mode size categories shown in Table 2. All ash samples used generally have a normal distribution for particle sizes, with the exception of the rhyolite sample from Chaitén, where particle sizes have a distinctive positive skew

Full size image

3.2.3 Estimated ash-settling rates

As the generation rate of each ash type and particle size is determined as part of the instrument calibration, estimations of settling rate can be made given the known volume of the container where the ash is dispersed and replicated under controlled conditions. As with particle sizes, all available worldwide ash-settling rate data were plotted, which showed a generally decreasing settling rate with distance from vent (Fig. 6).

Fig. 6
figure 6

Data from: 1Andò et al. (2014), 2Bonadonna et al. (2011), 3Bonadonna et al. (2002), 4Scott and McGimsey (1994), 5Harris and Rose (1983), 6Scheidegger and Federman (1982), 7Kienle and Swanson (1985), 8Brazier et al. (1982), 9Wilcox (1959). A value of 1100 g m−2 s−1 was also identified in Self et al. (1974), but the duration of this rate is unclear and it involved scoria material with a predominant particle size of ~ 6 mm. Therefore, we decided not to extrapolate it to 1 h for the purpose of this chart

Ash-settling rates and their associated distances from nine worldwide eruptive vents. The distances from central Auckland (depicted as 0 km) to New Zealand volcanic ash sources are identified by a series of dashed vertical lines, and approximate maximum axial extent of the AVF by a bold dashed vertical line.

Full size image

Based on the settling rates in Fig. 6, and the particle sizes adopted for experimentation (Table 2), which represent ashfall in Auckland from distal eruptions or mid- to lower-particle sizes expected from an AVF eruption, we attempt to replicate ash-settling rates between 50 and 1000 g m−2 h−1 (four specific ash-settling rates within this range per sample) for all six of the samples. However, we also attempt to replicate settling rates towards the upper limit of what can be expected in Auckland (i.e. around 2000 and 10,000 g m−2 h−1) for the Pupuke basalt and Chaitén rhyolite (category b) samples, where more material was available, to determine expected absolute minimum visual ranges from direct ashfall in the AVF. Thus, in addition to multiple trial tests, a total of 28 experiments are conducted as part of this study.

We highlight that the ash-settling rates determined at this stage are those forecast using the ash disperser calibration results. Actual ash-settling rates at the base of the container are lower; particularly for high ash-generation conditions, due to:

  • Adherence of some ash to the container side and top reducing the airborne concentration and settling rate.

  • An increase in pressure within the container acting against more ash entering as compressed air continues to enter at the top of the container, but the foam at the base becomes clogged.

  • Leakage of some ash from the container altogether, either through the foam or small gaps, especially when the air pressure inside is very high.

Actual ash-settling rates are determined from the centred mass balance and petri dish procedure outlined in Sect. 3.1.3.

4 Results

4.1 Ash concentrations and settling rates

Measurements of airborne ash concentration taken directly from the transmissometer were ~ 40–1600 mg m−3 for all experiments, with higher values corresponding to the highest ash-settling rates and lowest visual ranges (Online Resource 3–4). There were no distinct differences between airborne particle concentration reductions for the different ash samples or ash-generation rates we used in experimentation. It took 11 min (± 2 min) for airborne concentrations to return to original levels. However, given the exponential decay in airborne particle concentration following the cessation of new ash entering the container, values decreased to < 10% of their maximum values in less than 5 min.

For the relatively coarse-grained ash samples (particularly size category c, and to some extent b), a greater weight of ash and more ash of larger particle size fell towards the centre of the container base, with ash becoming more fine-grained towards the edge. Fine-grained ash adhered to the edge and top of the container (Fig. 7). Adherence was prevalent for the ash samples with the finest particle sizes (i.e. category a, and to some extent b), and especially so for the Chaitén rhyolite sample, causing sometimes large decreases in ash-settling rates towards the extreme edge of the base.

Fig. 7
figure 7

Annotated plan-view photographs taken looking down into the container showing a petri dishes on the central mass balance board and foam edge, and b describing general ash accumulation patterns across the container base, determined from the mass of ash on the centred board and in petri dishes, and through visual observations after each experiment

Full size image

Due to the inconsistencies across the container base, results displayed for ash-settling rates are in the form of three values, corrected for a 1-h period over 1 m2:

  1. 1.

    A minimum value calculated from the weight recorded on the centred mass balance board at the container base multiplied by a correction factor determined by the weight difference between ash in the petri dishes on the board and on the foam edges for that particular sample.

  2. 2.

    A maximum value calculated by doubling the weight recorded on the central mass balance board at the container base for that sample to assume the same weight falls on the foam edge.

  3. 3.

    The midpoint between points 1 and 2 above, which is deemed the value most likely to represent the true ash-settling rate.

Based on these values, although ash-settling rates forecast from the ash disperser calibration were ~ 50–10,000 g m−2 h−1 (Sect. 3.2.3), the actual ash-settling rates for our results are calculated at 28–4800 g m−2 h−1.

4.2 Visual ranges

Figure 8 shows an example of the direct visual range reading from the transmissometer for one experimental run. In our results, the maximum and minimum values recorded by the transmissometer have greater deviation from the mean for the lowest airborne concentrations and ash-settling rates.

Fig. 8
figure 8

Unprocessed transmissometer results for the Kaharoa rhyolite sample when ash was dispersed at the highest generation rate tested. Annotations show how the mean, maximum and minimum visual ranges were determined, as well as time taken for all ash to settle out and airborne concentration and visual range to return to original conditions

Full size image

For the Pupuke basalt, which had the highest bulk density of the four ash types used, testing was conducted using three distinct particle diameter size distributions, the results of which are summarised in Fig. 9 with data values in Online Resource 3. It is evident that for a given ash-settling rate, visual range is less for ash predominantly containing finer particles. Lines of best fit follow inverse power relationships (Fig. 9). Similarly, relationships between visual range and ash-settling rates are best described by inverse power law relationships for the rhyolitic and andesitic ash types (Fig. 10 and Online Resource 4 for data values). The Chaitén sample produced results with greater visual ranges than all other types in the same particle size category, despite being the lightest-coloured ash we tested.

Fig. 9
figure 9

Visual ranges for the three Pupuke basalt particle diameter size distributions. Each point is for the mean visual range when ash flow into the container was at equilibrium with ash-settling rate at the base for each experiment. Horizontal minimum and maximum error bar values and the point values correspond to bullet points 1–3 in Sect. 4.1, respectively. Vertical error bars show the maximum and minimum values recorded by the transmissometer during the periods of equilibrium (Fig. 8)

Full size image
Fig. 10
figure 10

Visual ranges for the four ash samples in particle diameter size distribution category b (i.e. 20–50 μm mode diameters). Each point is for the mean visual range when ash flow into the container was at equilibrium with ash-settling rate at the base for each experiment. Horizontal minimum and maximum error bar values and the point values correspond to bullet points 1–3 in Sect. 4.1, respectively. Vertical error bars show the maximum and minimum values recorded by the transmissometer during the periods of equilibrium (Fig. 8)

Full size image

5 Discussion

This section includes a discussion of our experimental approach and considerations for visibility in airborne volcanic ash in both a general sense and specifically forecast conditions for Auckland. We then outline key limitations of the approach adopted and recommendations for future adaptations and additions to the methodology. Finally, we consider implications of reduced visibility for transportation operation and provide examples of mitigation strategies that transportation and emergency management officials could implement based on findings from this study and other recent work.

5.1 Visibility in airborne volcanic ash

Our results demonstrate that the new methodology and experimental set-up developed is suitable to produce sufficiently consistent ash-settling rates to analyse visual range through airborne volcanic ash in laboratories. Compared to most other airborne particulate matter, ash particles are relatively large in size. However, their occurrence in large concentrations, like fog droplets, can cause substantial reductions in visibility, although they generally fall to the ground and out of suspension more quickly.

The time taken for the airborne ash concentration to reduce to zero upon cessation of dispersal at the top of the container is of particular interest. It provides an indication of how long remobilised ash would take to settle from a 1 m height above the ground given no ongoing disturbances such as wind or traffic and in the absence of rainfall. This type of information can be used to inform transportation management strategies such as the spacing of rail or road vehicles to allow sufficient time for ash to settle and reduce visual impairment to drivers (Sect. 5.3). Adherence of ash to the container is likely explained by the electrostatic charging of ash particles (Bonadonna and Phillips 2003; Folch 2012). However, as the light beam of the transmissometer passes the width of the container (1.4 m path length) and because inconsistencies occur in a radial pattern, this allowed the accurate calculation of mean ash-settling rates.

Visual range fluctuates between readings (Fig. 8), depending on the quantity and characteristics of particles within the light beam of the transmissometer at the precise time of measurement. However, as values are recorded at one-second intervals and because we calculate each mean visual range result from at least 30 min of experimentation, i.e. > 1800 data points, we consider our results reliable.

The result that visual range is lower for ash containing finer particles corresponds to the findings by Conner (1974), in that the mass scattering efficiency of light decreases as particle size increases. Comparing the Pupuke basalt and Kaharoa rhyolite, which were pulverised and sieved with the same dimension controls, results suggest that the light-coloured rhyolite causes lower visual ranges than the dark-coloured basalt. This makes intuitive sense, as there is more reflection from light-coloured surfaces. Low visual ranges for the andesite sample are perhaps explained by augite being one of the dominant minerals (constituting ~ 20% of the sample). With two prominent cleavages, meeting at angles near 90 degrees for augite, light scattering may be relatively high. Furthermore, the particle sizes for the andesitic ash are low compared to the basalt and Kaharoa rhyolite, and fallout of the andesite may be slowed by relatively irregular particle shapes (Riley et al. 2003; Pardini et al. 2016).

The most likely explanation for the unexpected visual range results obtained for the Chaitén sample is due to the larger positive particle size distribution from the mode than the other samples (Fig. 5), and thus higher proportion of relatively large diameter particles. Additionally, the comparatively low processing undertaken when producing this sample led to particles that are relatively spherical in shape (Fig. 11). As such, they reflect less light than the irregular-shaped particles of the other samples we tested. In fact, the Chaitén ash is more representative of fresh volcanic ash samples and our results for the other samples, which were all mechanically pulverised, thus provide a worst-case situation for visual range reduction during initial ashfall in this regard.

Fig. 11
figure 11

Microscope images for a Kaharoa rhyolite sample, b Chaitén rhyolite sample, both in particle size group b. The individual particles of the Chaitén sample are generally more spherical in shape. A Leica DM2500P microscope (with ×40 zoom) and Leica DFC295 camera attachment was used to capture the images

Full size image

5.1.1 Consequences of ashfall events in Auckland for visibility

Our results sufficiently cover the range of settling rates that can be expected in Auckland given a future eruption in New Zealand based on correlation with the limited dataset available from worldwide eruptions (Fig. 6). Ash-settling rates of ~ 1000 g m−2 h−1, possibly up to ~ 4000 g m−2 h−1, may occur in Auckland from any of these eruption locations (Fig. 6), and extrapolation of the line of best fit for the finest grained ash we tested (12 μm mode particle diameter) to these settling rates suggest that visual ranges as low as ~ 1–2 m can occur under such conditions.

Ash-settling rates are likely to be higher (resulting in shorter visual ranges) nearer to the vent and for explosive eruptions, which generally produce a relatively large proportion of fine-grained ash (White et al. 2011). For Auckland, small ash particle diameter sizes are possible from eruptions in both the AVF and TVZ of New Zealand (given appropriate wind directions). However, even the smallest estimated ash-settling rates in Auckland (~ 10 g m−2 h−1) during an eruption would cause a visibility reduction of over two hundred times from typical baseline visual range (i.e. < 300 m rather than 70 km visual ranges) for ash with characteristics in the range we tested.

5.2 Key limitations and suggestions for future work

  • Accurate source data for ash-settling rates (and particle concentrations) following past eruptions are somewhat limited, and forecasts remain constrained by the small dataset. We anticipate that the recent development of automated ash sampling instrumentation (e.g. Shimano et al. 2013; Weber et al. 2013, Andò et al. 2014; Tajima et al. 2015) and their deployment during and following future eruptions will improve understanding of settling rates and reduce uncertainty.

  • All of our experiments incorporated only fine-grained ash (all particle diameters < 320 µm with mode sizes < 110 µm) due to limitations of the ash dispersal equipment which was prone to clogging by coarse particles. However, the particle size distributions of our ash samples are generally polydisperse and appropriately replicate fine-grained ashfall that could occur in Auckland from diverse sources.

  • All ash samples were oven-dried to obtain consistent dryness prior to investigating particle characteristics. Therefore, we have not accounted for relative humidity affecting the ash over time, although the ash was exposed to relatively stable humidity in the laboratory when conducting the experiments. Visibility in mineral dust storms is known to decrease if the relative humidity is high, particularly if more than 70% (Hagen and Skidmore 1977). We speculate that if volcanic ash encounters higher relative humidity, there will likely be greater light scattering and lower visual range due to the hygroscopic properties that the particles exhibit (Malm et al. 2003; Barsotti et al. 2010). Therefore, Eq. 2 may underestimate visibility reduction.

  • Although the transmissometer records regular measurements of airborne particle concentration, we view these data with caution in this study, as precise values would require the density correction factor (set as 1.00 for our tests) to be adjusted between each sample. In the absence of complex isokinetic sampling of the airborne ash in the container, this cannot be achieved (DynOptic 2014), and we recommend this addition for future work. However, we adopted ash-settling rate as an alternative unit of measurement, which avoided the need for isokinetic sampling processes in our study.

  • It was evident that the Chaitén sample became highly electrostatically charged with many discharges occurring when touching the container both during experimentation and subsequent cleaning. More specifically, we suggest that the process of triboelectrification, described in relation to volcanic ash by Aplin et al. (2014) and Aplin et al. (2015), occurred frequently for the Chaitén sample. Triboelectrification is the type of contact electrification that occurs when two materials make contact with each other; Some electrical charge transfers from one material to the other with one gaining and the other losing electrons (Electrostatic Solutions Ltd. 2000). This process can occur readily in volcanic ash with a polydisperse particle size distribution dominated by very fine particle sizes (Aplin et al. 2014), such as our Chaitén sample (Fig. 5). If charged particles are physically separated, charges dissipate to earth, generating sudden electrostatic discharges. Additional testing using further samples is required to confirm this hypothesis and appropriately account for these electrostatic properties.

  • Given that ash deposits are often characterised across transects or accessible lines rather than across grids (Folch 2012), and given the number and complexity of variables associated with the remobilisation and resuspension of ash (e.g. meteorological conditions, ash wetness, particle characteristics, road surface specifics and vehicle type), we do not investigate the conditions for resuspended ash in detail. Rather, our calculations of visibility impairment largely represent conditions that can be expected during initial ashfall when ash accumulation on the ground is minimal. Current developments in resuspension modelling (e.g. Folch et al. 2014; Reckziegel et al. 2016; Miwa et al. 2018) will aid the future understanding of interactions between ash remobilisation and visibility effects. Our results are presently most reliable for road transportation where ground surfaces are wet and for maritime transportation, as atmospheric remobilisation will be minimal under these conditions. For ongoing eruptions however, our results should be treated as highly conservative for most land-based transportation types where cleaning has not occurred.

5.3 Considerations for transportation

From recent studies that examine impacts on road transportation from visibility degradation caused by fog (Sect. 2.6), we make comparisons to our experimental findings for airborne ash. It is evident that even the lowest ash-settling rates expected in Auckland and the largest particle diameter sizes (< 110 μm mode) used in our research would cause disruption to road transportation from speed reduction due to visibility degradation (< 600 m visual ranges). Speed reduction in volcanic ashfall may be even greater than for fog due to additional transportation impacts such as road marking coverage by ash and reduced skid resistance (Wilson et al. 2012a; Wilson et al. 2014; Blake et al. 2016b, 2017a). In the absence of key research on driver behaviour in volcanic ashfall and assuming no speed restrictions, based on data available for other atmospheric hazards we estimate an average free-flow speed of 30 km h−1 for a visual range of 5 m, with an increased chance of accidents due to lack of lane-keeping ability and obstructions hidden from view. With 20 m visual range, 50 km h−1 may be possible, but for the lowest expected visual ranges in Auckland (~ 1–2 m), driving is likely to become highly impractical and dangerous.

In an operational environment, transportation and emergency management officials may use knowledge of anticipated driver behaviour, along with results from our study and information on previous mitigation measures following historical eruptions (Table 1), to appropriately deal with reduced visibility (in addition to other impact types) from volcanic ash. For example, in dry conditions or during prolonged eruptions with recurring ashfalls, a 5-min spacing of road and rail vehicles may be appropriate as this allows ~ 90% of ash to settle (assuming no wind or other atmospheric disturbances). Additionally or alternatively, lower speed limits may be appropriate to minimise the resuspension of ash behind vehicles and reduce accident rates. Such mitigation measures align well with what was implemented on major routes following the 1980 Mount St Helens eruption; as Warrick (1981) states, “although the highways were officially closed, [vehicles] were allowed to proceed westward on I-90 and northward on US 97 at a rate of one vehicle every 5 min and a maximum speed of 25 mph” (p.21).

Transportation management policies and local regulations are likely to have a large influence on the level of disruption from volcanic ashfall in many locations. For example, in Auckland a “restricted visibility routine” for maritime area/s affected may come into effect when visibility of less than 1 nautical mile from a vessel’s bridge is encountered (AC 2014). This would have implications for the numbers and types of vessels permitted in certain areas, and their speeds (AC 2014; MNZ 2015). For airport operations at Auckland Airport, assuming no preceding closure due to other factors such as potential damage to aircraft engines, the visual ranges expected from most initial ashfalls mean that low visibility procedures would be initiated (triggered by < 1500 m visual range) (CAA 2008). Only essential ground support traffic directly involved with the arrival or departure of an aircraft will be allowed to operate in the manoeuvring area during such conditions. As the runway at Auckland Airport is equipped with a Category III B Instrument Landing System, aircraft and pilots with the appropriate ratings can land if the visual range is > 50 m (Auckland Airport 2008), with a maximum of six aircraft take-off and landing movements per hour (CAA 2008). With lower visual ranges, operations at the airport would cease, although closure would likely be necessary due to other factors by this stage.

6 Conclusions

We have developed an experimental approach to allow the calculation of visual range from given ash-settling rates for a range of different volcanic ash samples with different particle size and compositional characteristics. We have also tested this approach using a selection of volcanic ash samples that represent a range of characteristics that can be expected in Auckland, New Zealand, from multiple sources.

Our results demonstrate that visibility decreases with increasing ash-settling rate. Ash particle size has the greatest effect on visual range. For a given settling rate, fine-grained ash causes a shorter visual range than coarse-grained ash, a finding that aligns with existing research involving other particulate types. The influence of ash type, colour and shape on visual range reduction is not as clear-cut as for particle size. We conclude that samples with a higher proportion of irregular-shaped particles cause a greater reduction in visual range than those with more spherical shapes. Light-coloured ash such as rhyolite appears to result in lower visual ranges than dark-coloured ash such as basalt; a trend that may be explained on the basis of increased reflection off light-coloured surfaces. However, further laboratory or field-based studies are required to confirm the true extent that particle colouration and shape impacts visual range.

We emphasise that our results are for initial volcanic ashfall only and that any remobilised and resuspended ash from fall deposits could cause lower visual ranges for extended periods of time post-eruption, exacerbating impacts to transportation. Our study highlights the need for rapid syn-eruptive and reliable measurements or calculations of airborne metrics including particle concentration, ash-settling rate and visual range. It also demonstrates the importance of considering different ash properties besides ash thickness in volcanic eruptions, including particle size distribution, shape and colour.

Contemporary ash dispersion and fallout models, which provide outputs of airborne ash concentration and/or ash-settling rate given particular eruptions, can be used to produce scenarios of transportation disruption from reduced visibility, as well as other assessments including for human and animal health hazards. In conjunction with scenarios for other impact types (e.g. reduced skid resistance and line marking coverage by ash), the findings enable the improvement of transportation management strategies during and following future volcanic ashfall.