• Armienti, P., G. Macedonia, and M. T. Parsechi, 1988: A numerical model for the distribution of tephra transport and deposition: Applications to May 18, 1980 Mt. St. Helens eruption. J. Geophys. Res.,93, 6463–6476.

  • Casadavell, T. J., 1994: The 1989–1990 eruption of Redoubt Volcano, Alaska: Impacts on aircraft operations. J. Volcanol. Geotherm. Res.,62, 301–316.

  • Chen, C., and W. R. Cotton, 1983: A one-dimensional simulation of the stratocumulus-capped mixed layer. Bound.-Layer Meteor.,25, 289–321.

  • Chen, C., and W. R. Cotton, 1987: The physics of the marine stratocumulus-capped mixed layer. J. Atmos. Sci.,44, 2591–2977.

  • Cronin, S. J., M. J. Hedley, V. E. Neall, and R. G. Smith, 1998: Agronomic impact of tephra fallout from the 1995 and 1996 Ruapehu Volcano eruptions, New Zealand. Environ. Geol.,34, 21–30.

  • Deardorff, J. W., 1980: Stratocumulus-capped mixed layers derived from a three-dimensional model. Bound.-Layer Meteor.,18, 495–527.

  • Glaze, L. S., and S. Self, 1991: Ashfall dispersal for the 16 September 1986 eruption of Lascar, Chile, calculated by a turbulent diffusion model. Geophys. Res. Lett.,18, 1237–1240.

  • Glaze, L. S., S. M. Baloga, and L. Wilson, 1997: Transport of atmospheric water vapor by volcanic eruption columns. J. Geophys. Res.,102, 6099–6108.

  • Hurst, A. W., and R. W. Turner, 1999: Performance of a program for volcanic ashfall forecasting. N.Z. J. Geol. Geophys.,42, 615–622.

  • Lechner, P., 1997: Living with volcanic ash episodes in civil aviation. Civil Aviation Authority of New Zealand Tech. Note, 10 pp. [Available from Civil Aviation Authority of New Zealand, P.O. Box 31-441, Lower Hutt, New Zealand.].

  • Macedonio, G., M. T. Parsechi, and R. Santacroce, 1988: A numerical simulation of the Plinian fall phase of the 79 AD eruption of Vesuvius. J. Geophys. Res.,93, 14 817–14 827.

  • Macedonio, G., M. T. Parsechi, and R. Santacroce, 1990: Renewal of explosive activity at Vesuvius: Models for the expected tephra fallout. J. Volcanol. Geotherm. Res.,40, 327–342.

  • New Zealand Official Yearbook, 1997: New Zealand Official Yearbook of 1997. GP Publications, 645 pp.

  • Pielke, R. A., W. R. Cotton, C. J. Tremback, M. E. Nicholls, M. D. Moran, D. A. Wesley, T. J. Lee, and J. H. Copeland, 1992: A comprehensive meteorological modelling system—RAMS. Meteor. Atmos. Phys.,49, 69–91.

  • Pudykiewicz, J., 1988: Numerical simulation of the transport of radioactive cloud from the Chernobyl nuclear accident. Tellus,40B, 241–259.

  • Ritchie, H., and C. Beaudoin, 1994: Approximations and sensitivity experiments with a baroclinic semi-Lagrangian spectral model. Mon. Wea. Rev.,122, 2391–2399.

  • Searcy, C., K. Dean, and W. Stringer, 1998: PUFF: A high-resolution volcanic ash tracking model. J. Volcanol. Geotherm. Res.,80, 1–16.

  • Self, S., and G. P. L. Walker, 1991: Ash clouds: Characteristics of eruption columns. U.S. Geol. Surv. Bull.,2047, 81–86.

  • Sparks, R. S. J., M. I. Bursik, S. N. Carey, A. W. Woods, and J. S. Gilbert, 1991: The controls of eruption-column dynamics on the injection and mass loading of ash into the atmosphere. U.S. Geol. Surv. Bull.,2047, 81–86.

  • Suzuki, T., 1983: A theoretical model for dispersion of tephra. Arc Volcanism: Physics and Tectonics, D. Shimozuru and I. Yokoyama, Eds., Terra Publishing Co., 95–113.

  • Tremback, C. J., J. Powell, W. R. Cotton, and R. A. Pielke, 1987: The forward-in-time upstream advection scheme. Mon. Wea. Rev.,115, 540–555.

  • Tripoli, G. J., and W. R. Cotton, 1980: A numerical investigation of several factors contributing to the observed variable intensity of deep convection over south Florida. J. Appl. Meteor.,19, 1037–1063.

  • Tripoli, G. J., and W. R. Cotton, 1982: The Colorado State University three-dimensional cloud/mesoscale model—1982. Part I: General theoretical framework and sensitivity experiments. J. Rech. Atmos.,16, 185–220.

  • Vandemeulebrouck, J., A. W. Hurst, and N. Poussielgue, 1994: Implications for the thermal regime of acoustic noise measurements in Crater Lake Mount Ruapehu, New Zealand. Bull. Volcanol.,56, 493–501.

  • Walko, R. L., and C. J. Tremback, 1995: HYPACT: The Hybrid Particle and Concentration Transport Model. User’s guide, 13 pp. [Available from ASTER Division, Mission Reasearch Corporation, P.O. Box 466, Fort Collins, CO 80522.].

  • Wilson, C. J. N., 1991: Ash-fall deposits from large-scale phreatomagmatic volcanism: Limitations of available eruption-column models. U.S. Geol. Surv. Bull.,2047, 93–99.

  • View in gallery

    Isopachs (mm) for three tephra falls of the 1995 and 1996 Ruapehu event. [Figure adapted from Cronin et al. (1998)].

  • View in gallery

    Map showing the topography of the upper half of New Zealand (contour interval is 400 m) and the boundaries of the two inner RAMS grid (indicated by R2 and R3) and the boundary of the ASHFALL Model (indicated by ASH1). The other letters mark locations mentioned in the text as follows: R, Mount Ruapehu; A, Auckland; P, Paraparaumu; NP, New Plymouth; and G, Gisborne.

  • View in gallery

    Terminal velocity as a function of particle size (μm) as computed by the HYPACT model.

  • View in gallery

    National Oceanic and Atmospheric Administration visible satellite image for 1512–1516 NZST (=0312–0316 UTC) 17 Jun 1996. Notice the shadow cast by the ash cloud southeast of Lake Rotorua as indicated by the solid arrow. Note also the bright top of the eruption column as indicated by the dotted arrow.

  • View in gallery

    RAMS model–derived skew T soundings of temperature, dewpoint temperature, and winds for a point near Mount Ruapehu for (a) 1200 NZST 17 Jun 1996, (b) 0000 NZST 12 Oct 1995, and (c) 1800 NZST 14 Oct 1995. The labels on the vertical axis are pressure (hPa), the labels on the horizontal axis are temperature (°C), and the isotherms slope upward from left to right.

  • View in gallery

    RAMS/HYPACT-simulated distribution of settled ash particles at 2100 NZST 17 Jun 1996. The vertical source was assumed to be 3000 m thick, from 4000 to 7000 m in elevation.

  • View in gallery

    Same as Fig. 6, but the vertical source was assumed to be 500 m thick, from 6750 to 7250 m in elevation.

  • View in gallery

    RAMS/HYPACT-simulated (a) horizontal distribution of ash particles at 1516 NZST 17 Jun 1996 and (b) vertical distribution of ash particles at 1500 NZST 17 Jun 1996. The vertical source was assumed to be 500 m thick, from 6750 to 7250 m in elevation.

  • View in gallery

    Distance that ash of various particle sizes (μm) was transported downwind of Mount Ruapehu as computed by the HYPACT model using high-resolution RAMS-simulated winds for 17 Jun 1997.

  • View in gallery

    Isopach (mm) distribution for the 17 Jun eruption as simulated by the ASHFALL Model using fall velocity distribution 1 and the observed winds from the 1200 NZST 17 Jun rawinsonde launch. The location of Paraparaumu is marked by the P in Fig. 2.

  • View in gallery

    Isopach (mm) distribution for the 17 Jun eruption as simulated by the ASHFALL Model using fall velocity distribution 1 and the high-resolution RAMS-simulated winds for the location marked R in Fig. 2.

  • View in gallery

    RAMS/HYPACT-simulated distribution of settled ash particles at 0800 NZST 12 Oct 1995. The vertical source was assumed to be 3000 m thick, from 7000 to 10 000 m in elevation.

  • View in gallery

    RAMS/HYPACT-simulated distribution of ash particles above 1500 m elevation at 2300 NZST 11 Oct, and 0100 and 0500 NZST 12 Oct 1995. The orientation of the plume rotated clockwise with time. The vertical source was assumed to be 3000 m thick, from 7000 to 10 000 m in elevation.

  • View in gallery

    Isopach (mm) distribution for the 11 Oct 1995 eruption as simulated by the ASHFALL Model using fall velocity distribution 1 and the high-resolution RAMS-simulated winds for the location marked R in Fig. 2.

  • View in gallery

    RAMS/HYPACT-simulated distribution of (a) settled ash and (b) airborne ash particles at 0000 NZST 15 Oct 1995. The vertical source was assumed to be 3000 m thick, from 7000 to 10 000 m in elevation.

  • View in gallery

    Isopach (mm) distribution for the 14 Oct 1995 eruption as simulated by the ASHFALL Model using fall velocity distribution 1 and the high-resolution RAMS-simulated winds for the location marked R in Fig. 2.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 392 207 4
PDF Downloads 247 147 1

Factors Influencing Volcanic Ash Dispersal from the 1995 and 1996 Eruptions of Mount Ruapehu, New Zealand

View More View Less
  • a National Institute of Water and Atmospheric Research, Ltd., Wellington, New Zealand
  • | b Institute for Geological and Nuclear Sciences, Ltd., Lower Hutt, New Zealand
Full access

Abstract

The prediction of the dispersal of volcanic ash from events such as the Ruapehu eruptions of 1995 and 1996 is important, not only for civil-defense authorities who need to warn people in downwind areas, but for airline companies that have to reroute aircraft to avoid the encounters with volcanic ash clouds that can badly damage expensive jet engines and jeopardize passenger safety. The results of numerical simulations of volcanic ash dispersal using the Regional Atmospheric Modeling System (RAMS) and Hybrid Particle and Concentration Transport Model (HYPACT) for three periods (11–12 October 1995, 14 October 1995, and 17 June 1996) during the recent Ruapehu eruptive sequence are presented here. RAMS is a 3D atmospheric model that can be used to give detailed predictions of winds for regions such as the volcanic plateau. HYPACT is a particle dispersion model that uses the RAMS-generated wind fields to predict the movement and concentration of the volcanic ash cloud. Validation is achieved through comparison of predictions of airborne distributions and ashfall patterns with contour maps of ashfall depth and with satellite images of the ash cloud. Comparison of the performance of RAMS/HYPACT with that of the current Gaussian-plume “ASHFALL” Model currently used for ashfall advisory guidance in New Zealand indicates that the RAMS/HYPACT suite provides more accurate spatial and temporal forecasts than ASHFALL does, but that, like ASHFALL, its accuracy is limited by the accuracy of the initial and lateral boundary conditions provided and by the accuracy of the volcanological parameters that control the eruption-plume characteristics.

Corresponding author address: Richard W. Turner, National Institute of Water and Atmospheric Research, Ltd., P.O. Box 14901, Kilbirnie, Wellington, New Zealand.

r.turner@niwa.cri.nz

Abstract

The prediction of the dispersal of volcanic ash from events such as the Ruapehu eruptions of 1995 and 1996 is important, not only for civil-defense authorities who need to warn people in downwind areas, but for airline companies that have to reroute aircraft to avoid the encounters with volcanic ash clouds that can badly damage expensive jet engines and jeopardize passenger safety. The results of numerical simulations of volcanic ash dispersal using the Regional Atmospheric Modeling System (RAMS) and Hybrid Particle and Concentration Transport Model (HYPACT) for three periods (11–12 October 1995, 14 October 1995, and 17 June 1996) during the recent Ruapehu eruptive sequence are presented here. RAMS is a 3D atmospheric model that can be used to give detailed predictions of winds for regions such as the volcanic plateau. HYPACT is a particle dispersion model that uses the RAMS-generated wind fields to predict the movement and concentration of the volcanic ash cloud. Validation is achieved through comparison of predictions of airborne distributions and ashfall patterns with contour maps of ashfall depth and with satellite images of the ash cloud. Comparison of the performance of RAMS/HYPACT with that of the current Gaussian-plume “ASHFALL” Model currently used for ashfall advisory guidance in New Zealand indicates that the RAMS/HYPACT suite provides more accurate spatial and temporal forecasts than ASHFALL does, but that, like ASHFALL, its accuracy is limited by the accuracy of the initial and lateral boundary conditions provided and by the accuracy of the volcanological parameters that control the eruption-plume characteristics.

Corresponding author address: Richard W. Turner, National Institute of Water and Atmospheric Research, Ltd., P.O. Box 14901, Kilbirnie, Wellington, New Zealand.

r.turner@niwa.cri.nz

Introduction

The Mount Ruapehu (elevation 2900 m, latitude 39.289°S, longitude 175.562°E, location shown in Fig. 1) eruptions of 1995 and 1996 had an estimated cost to New Zealand’s economy of NZ$140 million (New Zealand Official Yearbook 1997). Contributing to this cost was that imposed on civil aviation because of the closure of airports by ashfall and the need to reroute aircraft because of the presence of ash clouds at flight levels. Aircraft need to avoid these clouds, because ash can accumulate on the combustor and remelted ash can accumulate on the inlet to the turbine section, which can lead to the engine stalling. In addition, the abrasive nature of the ash can damage the compressor blades and important areas of the exterior surface (Casadavell 1994). It is therefore important that volcanic ash advisories, which warn of the location and likely movement of ash clouds, are based on the most recent satellite information and accurate trajectory prediction methods to reduce the incidence of unnecessary (and costly) rerouting. In New Zealand, the Meteorological Service of New Zealand, Ltd., (MetService) produces “significant meteorology” advisories called volcanic ash “SIGMETs,” which form part of the interagency Volcanic Ash Advisory System (VAAS) (Lechner 1997). Other agencies involved in VAAS are the Civil Aviation Authority, the Airways Corporation of New Zealand, and the Institute of Geological and Nuclear Sciences (IGNS).

MetService’s VAAS responsibilities include, among others, maintaining “a watch over actual and possible volcanic events through the use of satellite and land-based meteorological systems and the use of atmospheric trajectory and dispersion models” and “us[ing] suitable atmospheric trajectory and dispersion models to identify the probable path of ejected ash . . .” (Lechner 1997). During the 1995 and 1996 eruptions, New Zealand’s MetService relied on the Volcanic Ash Advisory Center (VAAC) of the Canadian Meteorological Centre (CMC) for the production of the ash dispersal and trajectory forecasts. VAAC uses the three-dimensional Eulerian Canadian Emergency Response Model (CANERM), described in Pudykiewicz (1988), to forecast the medium- and long-term transport of volcanic ash. CANERM is generally operated with a horizontal spacing of 50 km and with 11 vertical levels (although the spacing can be halved if more detail is required). In forecast mode, CANERM uses meteorological fields provided by CMC’s global spectral model, which has a triangular 79-wave truncation that roughly corresponds to a horizontal spacing of 1.5° (Ritchie and Beaudoin 1994). Apart from the aviation-oriented CMC VAAC forecasts, during the 1995 and 1996 events IGNS produced for civil-defense purposes forecasts of volcanic ashfall with the “ASHFALL” Model (details of which are provided in section 2). Essentially, ASHFALL uses time-varying wind profiles at a single geographic location along with geological information such as the total volume of ash and distribution of fall velocities to compute ash deposition over the region downwind from the eruption.

Unfortunately, ASHFALL cannot account for any across-region variation in winds that may affect ash deposition. Additionally, the coarseness of CANERM’s horizontal grid means that orographic effects on the airflow such as topographic flow blocking and lee-wave generation that may affect its transport and deposition predictions may not be accounted for either. The Regional Atmospheric Modeling System [RAMS; see Pielke et al. (1992) and section 2a of this paper] can represent these kinds of effect for areas of complex terrain such as that of the Mount Ruapehu region (see Fig. 2) when configured with a horizontal grid spacing that is sufficiently fine to resove the important terrain features. Given the reliance of New Zealand’s VAAS system on ASHFALL and CANERM, it is important to quantify the effects of their limitations on forecast accuracy.

Because a high-resolution version of RAMS is capable of simulating the significant topographic effects, it is the main tool used in this study. The experimental design for the study involves using RAMS in both high-resolution [2.5-km horizontal grid spacing (Δx)] and low-resolution (40-km Δx) configurations to simulate the winds over the Mount Ruapehu region for three ash-dispersing periods during the 1995 and 1996 Ruapehu eruption episodes. The three periods were 11–12 October 1995, 14 October 1995, and 17 June 1996. These eruptions raised ash to heights of about 10 km above sea level and transported it hundreds of kilometers from the mountain as shown in Fig. 1. The isopach (lines of constant ash depth) maps of Fig. 1 (adapted from Cronin et al. 1998) are based on measurements along roads and on information provided by farmers.

The effect of model resolution on ash dispersion can be ascertained by comparing differences in the dispersion patterns produced when the Hybrid Particle and Concentration Transport Model (HYPACT) and/or ASHFALL use either the output from the coarse (Δx = 40 km) or high-resolution (Δx = 2.5 km) RAMS simulations. The effect of spatial variation in the winds can be identified through comparing the difference in ash deposition between HYPACT and ASHFALL.

The Ruapehu dataset also provided an opportunity to calibrate the RAMS and HYPACT models in the New Zealand region. Validation of the model results in this study is achieved by comparison with satellite images, radar data, ashfall data, and eyewitness reports. Direct validation of RAMS upper-level winds is not possible, because there are no upper-air observations in the region. In addition, the dataset allowed estimates of the maximum eruption-column height made by the volcanic plume entrainment model of Glaze et al. (1997) to be checked for their operational usefulness.

The dispersion models in this study are not appropriate tools for prediction of volcanic ejecta such as hot gases and large ballistic rocks, which are hazards that occur close (i.e., within a few kilometers) to the volcano. For any major volcanic eruption, it is assumed that the immediate vicinity of the volcano would be evacuated and the air space above be declared a no-fly zone. The dispersion models are most applicable to “sub-Plinian” or “Plinian” eruptions in which a large quantity of ash is generated (typical of the New Zealand andesite volcanoes Ruapehu, Ngauruhoe, and White Island). For very large eruptions, such as that of 26 000 years ago that formed Lake Taupo (Wilson 1991), the “umbrella” portion of the ash cloud can propagate considerable distances upstream against the prevailing wind. In these cases the dispersion models may be limited in their applicability, because one could only simulate the ash dispersal once it had moved some distance away from the volcano.

The paper is organized as follows: section 2 provides a description of the models used in this study, section 3 outlines the experimental design, the results are presented in section 4, and conclusions along with a discussion are presented in section 5.

Models

A brief description of the four numerical models used in the study now follows.

RAMS

Version 3b of RAMS was used for this study. The grid configuration used within RAMS consisted of three nested Arakawa C grids. The outer grid (grid 1) was a 44 × 52 grid with a horizontal spacing of 40 km and a time step of 100 s. The first nested grid (grid 2) was a 62 × 62 grid with a horizontal resolution of l0 km and a time step of 25 s. The innermost grid (grid 3) centerd over Mount Ruapehu was a 70 × 86 grid with a horizontal resolution of 2.5 km and a timestep of 5 s. The boundaries of the two inner RAMS grids are shown in Fig. 2. Boundary forcing was provided by the twice-daily 2.5° analyses from the European Centre for Medium-Range Weather Forecasts. The vertical grid structure had 22 vertical levels with a spacing of 200 m near the surface, increasing to 800 m above 2400 m. The vertical grid structure was the same for all three grids.

The computational expense involved with the 3-grid configuration is high. However, using this configuration enables the influence of relatively finescale topographic flows on ash dispersal and deposition to be determined, and thus the benefits of running an operational version of the RAMS model in this configuration (for the purposes of providing accurate meteorological input to HYPACT or ASHFALL) can then be assessed.

The following options and assumptions regarding the model’s configuration were invoked: 1) the basic equations were compressible and nonhydrostatic as in Tripoli and Cotton (1980), 2) the time-differencing scheme was a hybrid of two schemes in which velocity components and pressure are updated using the leapfrog scheme as described in Tripoli and Cotton (1982) and all other variables are updated using the forward-difference scheme described in Tremback et al. (1987), 3) a second-order advection scheme was used, 4) a prognostic equation for turbulent kinetic energy (TKE) for turbulence closure as in Deardorff (1980) was used, 5) the vertical coordinate was a terrain-following sigma coordinate, 6) no cumulus parameterization or microphysical packages were used, 7) longwave and shortwave radiation packages as described in Chen and Cotton (1983, 1987) were used, 8) a rigid lid at 18.5 km was specified, 9) zero-gradient inflow and an outflow eight grid points deep at the lateral boundaries were specified, 10) roughness length of 0.1 m was assigned, 11) an 11-level, sandy loam, soil model was used, 12) monthly mean SST were used, and 13) mixed woodland land use was specified.

HYPACT

The HYPACT model (Walko and Tremback 1995) simulates the motion of atmospheric tracers under the influence of winds and turbulence. Its Lagrangian component enables representation of sources of any size and the maintenance of concentrated, narrow plumes until atmospheric dispersion dictates that they should broaden. At this point, the Lagrangian particle plume can be converted into a concentration field and then advected using an Eulerian formulation. The Lagrangian particles are moved through space and time based on the interpolated wind velocities plus a superimposed random motion scaled on the local turbulent intensity. In addition, a spectrum of gravitational settling velocities related to particle size can be specified. Along wind, crosswind, and vertical wind components (u, υ, w); potential temperature (θ), and TKE are the gridded meteorological variables necessary to drive HYPACT. Gridded time series of these variables were provided by the RAMS model.

For this study, the Eulerian mode of HYPACT was configured to have a horizontal grid spacing of 4 km. Particles representing a volume of volcanic ash of 25 000 m3 each were released at a rate of 20 per time step of 100 s. This value gives a volume of approximately 0.1 km3 for a 6-h eruption of constant emission rate, which corresponds to an average depth of 1 mm over an area of 100 000 km2. The particles were assumed to be between 1 and 200 μm in diameter with an equal probability of any diameter particle being emitted between these limits. Note that, for an actual eruption, the particle size distribution may not fit this assumption. The terminal fall velocity for particles is approximated by Stokes law in a manner similar to Searcy et al. (1998) and is a function of particle size as shown in Fig. 3. The Stokes formulation is not a good approximation for particles with diameters exceeding 80 μm; however, using it for larger size particles allows for an easy comparison with the ASHFALL Model where fall velocities, rather than particle diameters, are directly specified.

ASHFALL

The ASHFALL program (described in Hurst and Turner 1999), which calculates ash thickness, was developed from the volcanic ash dispersion model described in Macedonio et al. (1988, 1990) and Armienti et al. (1988). Their “DIFFUSE” Model performs computationally expensive 3D particle-diffusion calculations for volcanic ash clouds and is unsuitable for civil-defense purposes. ASHFALL is an adaptation of DIFFUSE in which vertical diffusion is eliminated. This simplification eliminates the need for large 3D arrays, because, at any stage of the numerical integration, calculations of ash dispersal are only being made at a single atmospheric level. It has been argued that this assumption is reasonable given that the vertical diffusion of ash is always less important than horizontal diffusion and its effect can be crudely accounted for by increasing the horizontal diffusion coefficient (Suzuki 1983; Glaze and Self 1991).

ASHFALL uses a time-varying vertical profile of wind speed and direction for a location usually near the volcano (i.e., ASHFALL can account for vertical, but not horizontal, variation in winds) along with volcanological information such as the total volume of ash and distribution of fall velocities to calculate the likely distribution of ash thickness resulting from a volcanic eruption. ASHFALL does not consider the effect of lahars, pyroclastic flows, volcanic bombs, and gas clouds. Although these phenomena represent a considerable hazard close to the volcano, these areas are likely to be evacuated no matter which direction the ash is expected to travel.

ASHFALL was configured to have an 81 (E–W) × 121 (N–S) grid with a horizontal spacing of 5 km and a vertical spacing of 500 m up to a height of 12 km. The boundary of the ASHFALL grid is shown in Fig. 2. In most cases, a fall-velocity distribution as given in column 4 of Table 1 was assumed for the ash particles. The time-varying wind profiles were from RAMS winds interpolated to locations upstream of Mount Ruapehu. Some sensitivity tests as to the specified fall-velocity distribution were conducted and are reported later in the paper.

Plume-rise model

As mentioned in the introduction, attempts at estimating the height of the eruption columns were made using the plume entrainment model described in Glaze et al. (1997). Apart from determining the redistribution of atmospheric water by the eruption column by solving the equations of conservation for mass, momentum, and heat for the four separate components of the eruption column (dry air, water vapor, liquid condensates, and solid particles), this model determines the level of neutral buoyancy (approximately the level at which the density of the column equals the density of the environment) and the final eruption height (the level at which the vertical velocity of the eruption column reaches zero). Input variables to the model are profiles of atmospheric temperature and moisture, initial upward velocity of the plume (u0), initial temperature of the plume (θ0), initial plume radius (r0), and the initial mass fraction of water vapor (n0).

The structure of an eruption column can be briefly described as follows [for more details the reader is referred to Self and Walker (1991) and Sparks et al. (1991)]. The eruption column consists of a lowermost gas-thrust region, a convective thrust region, and an uppermost umbrella region. The gas-thrust region at the base of the eruption column is a jet produced by a decompression-induced expansion of the eruptive mixture as it encounters lower atmospheric pressure at the mouth of the vent. The material within this jet is initially denser than the ambient air, and exit velocities are typically in the range of 100–500 m s−1. The flow in the gas-thrust region, although complex, makes up only a small fraction of the total column height and generally has little influence on ash dispersal; the exception is the scenario in which the column collapses. For most eruptions, turbulent entrainment of the environmental air into the column and its subsequent heating causes the column to become positively buoyant, and above this point (typically a few hundred meters in elevation) the eruption column can be thought of as a convective plume. In some instances, the entrainment and heating of the air are insufficient for the column to reach a state of positive buoyancy; in this case, the column collapses, and a pyroclastic gravity current that flows close to the terrain is generated. In this instance, dispersion models are inappropriate tools for predicting ashfall dispersal, but ash will usually only travel a few kilometers from the volcano.

The upper limit of the convective region of the column is that level at which the plume loses its positive buoyancy and becomes first neutrally and then negatively buoyant (this level is called the level of neutral buoyancy). Above the level of neutral buoyancy, the column will still rise because of its upward momentum, and it will continue to do so until it reaches its maximum height when vertical velocity reaches zero. Between the level of neutral buoyancy and the top of the column, the ash plume will spread out laterally because vertical displacements are limited; this region is called the umbrella region and is analogous to the anvil of a cumulonimbus thunderstorm cloud. The transport of ash into the umbrella region is very efficient, because there is very little ash fallout from the gas-thrust and convective regions and thus it is the wind dispersal of the ash (along with gravitational settling) from the umbrella region of the eruption column that is most important for civil aviation interests and for determining longrange ashfall distributions.

Results

In this section the results of the simulations using the RAMS-driven HYPACT and ASHFALL Models of ashfall dispersal from three periods of the 1995 and 1996 Ruapehu eruption sequence are presented separately.

17 June 1996

The major ash-producing eruption of June 1996 began on the morning of 17 June at 0700 NZST (1900 UTC 16 June). Observer reports indicate that the activity was characterized by two eruption pulses at 0710 and 0825 NZST and steady activity from 1100 up until 1300 NZST, after which a decline occurred. At about 1500 NZST, the volcano started to erupt every 10–15 min, with ash-laden plumes reaching several kilometers in altitude. This activity had declined by 1700 NZST. There was another period of strong tremors with lava fountaining between 2200 and 0300 NZST that night. Ash, associated with lava fountaining, tends to be injected at low levels, and winds in the lee of Mount Ruapehu at levels below 1.5 km were modeled to be blowing to the northwest at this time. This result may explain the minor amount of ash observed to have settled in the northwest-oriented band (Fig. 1). However, the fact that major ashfalls were associated with the ash injections earlier in the day is supported by eyewitness reports. These reports indicated that large ashfalls occurred in a zone extending north-northeast (N-NE) from the volcano to the Bay of Plenty coast between Tauranga and Whakatane (New Zealand Official Yearbook 1997). The winds over the North Island on this date were generally south to southwesterly aloft with a trend to more southerly flow at lower levels. The 1514 NZST visible satellite image (Fig. 4) shows clear skies over most of the North Island (with the exception of the ash cloud). The 1514 NZST infrared satellite image (not shown) showed the ash cloud extending from its source at Mount Ruapehu toward Lakes Taupo and Rotorua and then spreading out transversally as it went N-NE out to sea. This infrared image indicated that the temperatures along the top of the ash plume’s centerline were between 235 and 239 K. Analysis of the RAMS simulation at 1500 NZST (see Fig. 5a) indicates that these temperatures correspond to a height above sea level of between 6.7 and 7.4 km (or between 430 and 400 hPa). Using the visible satellite image and the solar declination angle for this location and time (17°), it can be determined from the shadows cast by the edge of the ash cloud that that part of the cloud is between 6 and 7 km high. Glaze et al.’s theoretical plume model using typical volcano values of τ, u0, θ0, and r0 (see Table 2); a vent height of 2531 m (Vandemeulebrouck et al. 1994); and the RAMS sounding yielded values between 8.5 and 11.5 km for the final plume height. The calculated final plume height displayed little sensitivity (less than 1%) to the details of the tropospheric sounding, as shown when the warmer and moister soundings from the 11 and 14 October 1995 cases (see Figs. 5b and c, respectively, for these soundings) were used as input. Table 2 summarizes these sensitivity tests and shows that the main determinants of plume height are the volcano emission parameters.

Given the disagreement between the three estimates of the height of the plume top, a number of heights between these limits of 6.5 and 11.5 km were specified in HYPACT. The specified plume-top height that produced the best simulated match to the ashfall reports and satellite image turned out to be 7 km; the worst match was for 11 km. The best fit between simulated and observed ashfall was achieved with HYPACT when the ash cloud above the volcano was assumed to be 3 km deep and to have a top at 7 km (i.e., the source was from 4 km to 7 km). The results of this simulation are shown in Fig. 6, which indicates (by comparison with Fig. 1) that the simulated ashfall pattern has the correct orientation but is displaced about 15 km too far to the southeast near Rotorua. Figure 7 shows the HYPACT-simulated ashfall when the source has a much smaller vertical extent of 500 m (from 6500 to 7000 m elevation); here the southeast displacement near Rotorua is about 25 km. Note that, for the simulations shown in Figs. 6 and 7, ash was assumed to be continuously ejected from 0700 to 1700 NZST on 17 June, and the plots, which are valid for 2100 NZST (4 h after the injection has stopped), show only those particles that have fallen below 1.5-km elevation. Figure 8a shows the HYPACT-simulated ash particles that are still suspended in the air and are above 1.5-km elevation at 0315 NZST on 17 June 1996. In comparing this figure with Fig. 3 it is apparent that the plume is again displaced about 30 km too far to the southeast over the Bay of Plenty coast. This displacement increases to about 70 km at 36°S, 178°E. Figure 8b shows the simulated vertical distribution of ash particles at 0300 NZST; the high concentrations at ground level nearer the volcano are readily apparent. This plot reveals one difficulty with HYPACT in deriving quantitative ashfall amounts in that little ash is simulated to have fallen in the vicinity of the volcano. A possible remedy to this problem would be to skew the specified distribution of particles within HYPACT toward those with larger diameters (i.e., >200 μm); however, the Stokes law used for computing fall speeds is a very poor approximation for these size particles, and unrealistic low-level dispersion patterns were simulated when large particle sizes were specified. Different-sized ash particles were simulated by HYPACT to settle at different distances from the volcano for this case and all the other cases. Figure 9 shows that, for this case, particles with diameters greater than 100 μm were simulated to be transported less than 200 km downwind, but the simulated transport distance of particles of diameter less than 50 μm was greater than 500 km.

The results from the plume dispersion simulations discussed thus far have been for HYPACT forced with output from high-resolution (i.e., 2.5-km horizontal spacing on the inner grid) RAMS integrations. Use of the coarser 40-km-spacing RAMS output (not shown) resulted in similar upper-level patterns being simulated. Low-level patterns were somewhat different, with the main axis of the plume being rotated slightly to the northwest for the 40-km simulation. However, for other dates for which no eruptions occurred but RAMS, in high-resolution mode, has forecast strong lee waves, differences in the vertical position of hypothetical ash plumes of up to 1.5 km have been simulated. For these same cases, differences of up to 10° in the horizontal orientation of the hypothetical plume axis in comparison with the-low resolution RAMS output have been simulated.

Ashfall patterns as predicted by ASHFALL with wind profiles provided by the 1200 NZST 17 June 1996 Paraparaumu rawinsonde and high-resolution RAMS simulation are given in Figs. 10 and 11, respectively. (The distribution of fall velocities for these simulations is given in column 1 of Table 2.) It can be seen that both patterns are somewhat similar, with the Paraparaumu-forced simulation being better (in comparison with the distributions in Fig. 1) in that it has captured the orientation of the plume more accurately. Both overestimate the extent of the 1-mm isopach, but this result is likely due to the fact that the value of 0.1 km3 for total volume of ash ejected was just a rough estimate. Although the Paraparaumu rawinsonde winds were representative of the conditions over the volcanic plateau on this day (in which the whole North Island was embedded in a south-southwesterly airstream and Paraparumu was upstream of Ruapehu), this case may not always be true, and reliance on remote, albeit observed, wind data may not always produce such good results. To illustrate this point, upper-air winds from New Plymouth for 1200 NZST 17 June were also used to force ASHFALL and produced an ashfall distribution (not shown) that was even more displaced to the southeast than was the RAMS-forced distribution. Using the fall velocity distribution as specified in column 2 of Table 1 resulted in relatively minor differences in the ashfall distribution obtained when column 1 was used, indicating that ASHFALL shows little sensitivity to these specified fall velocity distributions. In contrast, HYPACT simulations using particle size distributions that mimic these fall velocity distributions showed differences of about 150 km in how far the closest ashfall to the volcano was. The reason for this difference is that ash is assumed to be lost from all altitudes within the eruption column in ASHFALL.

11 October 1995

Eyewitness (army personnel and pilots) reports indicate that seismic activity between 2100 NZST 11 October and 0500 NZST 12 October were accompanied by a continuous explosive eruption that produced an eruption column between 8 and 10 km high. These height values are in better agreement with that simulated by the plume model for this date (see Table 2) than for the 17 June 1996 case. Winds were generally from the southwest at the beginning but then switched more to the west by the end of the period. This switch, along with the fact that the amount of ash from this eruption (about 0.2 km3) was the largest since 1945, meant that a much larger area was affected by ashfall than for the other two cases in this study (Fig. 1). No satellite imagery was available, and eyewitness reports indicated that the weather conditions at this time were not conducive to good observations of the ash cloud.

For the dispersion simulations on this date, HYPACT was configured to have a source 3-km deep centerd around 8.5 km, and ash particles were released continuously for the 8-h period from 2100 NZST 11 October to 0500 NZST 12 October. The simulation was stopped at 0800 12 October. The ash particles that had settled to below 1500 m are plotted in Fig. 12. From this figure, it is clear that the HYPACT simulation has produced an ashfall pattern that is slightly more east–west oriented than was observed. HYPACT has not simulated the trace amounts that fell in the north near Opotiki. However, HYPACT did simulate the movement of the elevated part of the ash cloud over this region and its subsequent movement to the south as the wind switched to the west (Fig. 13).

The ashfall pattern as predicted by ASHFALL with wind profiles provided by the high-resolution RAMS simulation between 2100 and 0800 NZST for the grid location upstream of Ruapehu is given in Fig. 14. The extent of the 1-mm isopach from the volcano is well simulated, suggesting the estimate of the volume of volcanic ash of 0.2 km3 for this event is reasonable. However, ASHFALL fails to capture the falls around Wairoa and, like HYPACT, fails to produce the ash that fell to the north. In fact, ASHFALL gives no indication that the ash cloud ever got close to Opotiki.

14 October 1995

Seismic reports for this event indicated much stronger tremors than what had occurred on the 11 October; however, the ash-producing eruption was reported to have lasted for only 4 h between 1530 and 1930 NZST on 14 October, and the volume of ash ejected was estimated to have been 0.1 km3. Ash plumes were reported to have risen as high as 11 km, and, at 2000 NZST, satellite imagery (not shown) indicated the ash cloud at an elevation of 10 km extended over the coast between Napier and Waipukurau. As for the previous case, the value for the height of the plume top is again in reasonable agreement with that of the plume model. Winds for this period were strong and steady from the north-northwest. Eyewitness reports indicated that the weather conditions were not good for measuring ashfall. However, enough ground observations were made to produce the distribution as shown in Fig. 1.

For the dispersion simulations on this date, HYPACT was configured to have a source 3-km deep centerd around 8.5 km, and ash particles were released continuously for the 4-h period from 1530 to 1930 NZST on 14 October. The simulation was stopped at midnight. The distribution of ash particles that had settled to below 1500 m is shown in Fig. 15a. From this figure, it is clear that the HYPACT simulation has produced an ashfall pattern that is slightly more oriented to the southeast than that reported in Fig. 1. The orientation of the elevated (i.e., above 1.5 km) part of the ash cloud (at 2000 NZST) agrees well with the description from the satellite image. Note that the low-level ash is displaced to the southeast of the upper-level ash (Fig. 15b), suggesting that the ash particles encountered more-northerly winds as they fell to the surface. However, the indications from Fig. 1 are that the bulk of the ash fell directly beneath the elevated plume. In fact, it seems to have been the case for all three RAMS/HYPACT simulations that the centerline of the simulated ashfall is rotated clockwise by about 10° when compared with the observations. This was not the case for the airborne ash at upper levels, for which the simulations were more accurate. One possible reason for this difference is that the degree of frictional turning within the boundary layer is too great in the RAMS results. Another possibility is that there are internal ash-cloud dynamics or there are interactions between the ash cloud and the atmospheric environment that are not being accounted for in the model. For example, the strong shading by the ash cloud on the ground could have been sufficient to alter the properties and flow within the planetary boundary layer beneath the cloud, although this latter mechanism cannot explain the errors in the nighttime forecasts for the 14 October case.

The ashfall pattern as simulated by ASHFALL with wind profiles provided by the high-resolution RAMS-simulated winds that are valid hourly between 1530 and 2330 NZST on 14 October for the grid location upstream of Ruapehu is given in Fig. 16. The extent of the 1-mm isopach from the volcano is oversimulated, much like that for the 17 June case, again suggesting the estimate of the volume of volcanic ash for this event of 0.1 km3 is too high. The orientation of the plume has the same kind of error as that for HYPACT, in that the center of the plume is about 20 km too far to the south at the coast. Given that the distance from this part of the coast to Mount Ruapheu is about 120 km, this displacement corresponds to an error in simulated wind direction of about 10°.

Discussion and conclusions

The results of numerical simulations of volcanic ash dispersal using the RAMS meteorological model wind–driven HYPACT and ASHFALL dispersion models for the 1995 and 1996 Mount Ruapehu eruptions have been presented and are now discussed. The most important factors in predicting airborne ash concentration location and ash deposition for New Zealand are volcanic emission parameters such as vent radius, initial upward velocity of the plume, initial plume temperature, and the total volume of ash ejected into the atmosphere. The first four parameters listed controlled the height and depth of the layer from which the ash is dispersed by the winds and are important because the winds that transport the ash can vary greatly in speed and direction as the altitude changes. Quantitatively, the airborne concentration and depths of ash deposition were most obviously dependent on the total volume of ash ejected. Details of the tropospheric temperature and humidity structure had little effect on the height to which the eruption column reached according to the plume model of Glaze et al. (1997). This model produced good simulations of the maximum plume height reached in the October 1995 cases but overestimated the maximum plume height for the June 1996 case. Given the lack of accurate knowledge about the initial plume characteristics, for emergency response purposes it would be best to generate a number of ash dispersal forecasts assuming a wide range of eruption column heights. Table 3 illustrates this point. For each of the three cases, a range of different ejection heights and eruption times in HYPACT was specified, and, after 8 h, the mean orientation of the plume centerline within 200 km of the volcano was determined. It is seen from the table that, for these cases, an envelope of approximately 35° would have been be a useful range to place on any predictions of ashfall given an assumption of accurate meteorological input but with allowing for some uncertainty about the strength and timing of the eruption.

Ash-particle size distributions also had an effect on the simulated deposition patterns. For the cases in this paper, skewing the distribution to lighter particles (radii less than 50 μm) with smaller fall velocities resulted in only very light amounts of ash being deposited and only at considerable distances from the source; skewing the distribution toward particles with large radii (>150 μm) resulted in no ash being transported more than 100 km from the volcano. An assumption of having an even distribution of particle sizes between 10 and 150 μm produced patterns that matched the observations reasonably well for distances greater than 100 km from the volcano.

Specifying the correct time of the main ash eruptions was also important, because wind direction and speed can change significantly over the course of a few hours. This effect was demonstrated for the 11 October 1995 case for which a forecast simulation with an assumed start time for the eruption of 3 h after what actually occurred would have had positional errors of up to 100 km at a distance of 200 km from the volcano because of a wind shift from the southwest to the west. Satellite imagery could be used when and where the temporal coverage is adequate to provide this timing, as well as the initial plume position and elevation information.

Flow features that are more accurately simulated within the 2.5-km model domain, such as lee waves, could have produced differences in the vertical position of the plume of up to 1.5 km in comparison with a 40-km-resolution simulation. Differences in the horizontal orientation of the plume axis of up to 10° could also have occurred between the high- and low-resolution simulations. Unfortunately, the gross scale nature of the observed data made it difficult to assess the quality of the high-resolution RAMS simulations. Given this fact and the fact that there were no other large-scale differences between the low-resolution (40 km) and the high-resolution (2.5 km) RAMS simulations, it would seem that, for operational purposes, the low-resolution RAMS model could be used to provide forcing of forecast data to either ASHFALL or HYPACT to produce adequate forecast guidance. Nevertheless, high-resolution spatially varying estimates of wind speed and direction are important enough to warrant their incorporation into a volcanic ash dispersion forecast system when computationally practicable.

The HYPACT and ASHFALL Models have been shown to produce reasonable, but not exact, agreement with the observed ashfall patterns for the three major ash-producing eruptions of the 1995 and 1996 Ruapehu eruption sequence. The HYPACT model is superior in reproducing the temporal and spatial movement of the ash cloud but, because of limitations in how the distribution of fall velocities are computed, is currently not suitable for quantifying the depth of ash. However, for prediction of the airborne movement of ash for civil aviation purposes, HYPACT was superior, as was demonstrated by its performance for all three cases examined in this study.

Both HYPACT and ASHFALL run very quickly, and scripts can be easily set up so that local ash dispersion pattern forecasts could be made using both models within 10 min of a request for a forecast, provided the appropriate numerical wind forecast is easily accessible. (In certain instances, when observed wind profiles measured at locations such as Paraparaumu are available and these profiles are representative of the wind near the volcano, it may be more appropriate to use these observations as input to the models rather than the numerical wind forecasts.) Procedures could easily be set up whereby a weather forecaster or geologist armed with some simple instructions could produce the ashfall forecasts, without the assistance of the numerical modeler. One advantage of this method would be that ashfall patterns from a number of different eruption scenarios could be generated by varying inputs to HYPACT and/or ASHFALL such as the height of the ash cloud and the time of the eruption.

Acknowledgments

We thank the following individuals and organizations for their assistance in this research: Peter Lechner of the Civil Aviation Administration and Kip Marks, Mark Sinclair, David Wratt, Mike Uddstrom, Neil Gimson, and Martin Bell of New Zealand’s National Institute for Water and Atmospheric Research (NIWA) for assistance in gathering data, satellite images, and providing advice on using the RAMS and HYPACT models; Shane Cronin of Massey University for the ashfall data; Alan Seed of University of Auckland for a description of the radar images of the ash cloud; and Lori Glaze of NASA Goddard for the plume model. The first author thanks NIWA and the Foundation for Research Science and Technology (FRST) for supporting this work under the auspices of the FRST postdoctoral fellowship scheme under Contract C01638. The second author thanks the Earthquake Commission Research Foundation for support under Contract EQC 97/268. Last, we thank the three anonymous reviewers whose helpful comments and criticisms improved the quality of the paper.

REFERENCES

  • Armienti, P., G. Macedonia, and M. T. Parsechi, 1988: A numerical model for the distribution of tephra transport and deposition: Applications to May 18, 1980 Mt. St. Helens eruption. J. Geophys. Res.,93, 6463–6476.

  • Casadavell, T. J., 1994: The 1989–1990 eruption of Redoubt Volcano, Alaska: Impacts on aircraft operations. J. Volcanol. Geotherm. Res.,62, 301–316.

  • Chen, C., and W. R. Cotton, 1983: A one-dimensional simulation of the stratocumulus-capped mixed layer. Bound.-Layer Meteor.,25, 289–321.

  • Chen, C., and W. R. Cotton, 1987: The physics of the marine stratocumulus-capped mixed layer. J. Atmos. Sci.,44, 2591–2977.

  • Cronin, S. J., M. J. Hedley, V. E. Neall, and R. G. Smith, 1998: Agronomic impact of tephra fallout from the 1995 and 1996 Ruapehu Volcano eruptions, New Zealand. Environ. Geol.,34, 21–30.

  • Deardorff, J. W., 1980: Stratocumulus-capped mixed layers derived from a three-dimensional model. Bound.-Layer Meteor.,18, 495–527.

  • Glaze, L. S., and S. Self, 1991: Ashfall dispersal for the 16 September 1986 eruption of Lascar, Chile, calculated by a turbulent diffusion model. Geophys. Res. Lett.,18, 1237–1240.

  • Glaze, L. S., S. M. Baloga, and L. Wilson, 1997: Transport of atmospheric water vapor by volcanic eruption columns. J. Geophys. Res.,102, 6099–6108.

  • Hurst, A. W., and R. W. Turner, 1999: Performance of a program for volcanic ashfall forecasting. N.Z. J. Geol. Geophys.,42, 615–622.

  • Lechner, P., 1997: Living with volcanic ash episodes in civil aviation. Civil Aviation Authority of New Zealand Tech. Note, 10 pp. [Available from Civil Aviation Authority of New Zealand, P.O. Box 31-441, Lower Hutt, New Zealand.].

  • Macedonio, G., M. T. Parsechi, and R. Santacroce, 1988: A numerical simulation of the Plinian fall phase of the 79 AD eruption of Vesuvius. J. Geophys. Res.,93, 14 817–14 827.

  • Macedonio, G., M. T. Parsechi, and R. Santacroce, 1990: Renewal of explosive activity at Vesuvius: Models for the expected tephra fallout. J. Volcanol. Geotherm. Res.,40, 327–342.

  • New Zealand Official Yearbook, 1997: New Zealand Official Yearbook of 1997. GP Publications, 645 pp.

  • Pielke, R. A., W. R. Cotton, C. J. Tremback, M. E. Nicholls, M. D. Moran, D. A. Wesley, T. J. Lee, and J. H. Copeland, 1992: A comprehensive meteorological modelling system—RAMS. Meteor. Atmos. Phys.,49, 69–91.

  • Pudykiewicz, J., 1988: Numerical simulation of the transport of radioactive cloud from the Chernobyl nuclear accident. Tellus,40B, 241–259.

  • Ritchie, H., and C. Beaudoin, 1994: Approximations and sensitivity experiments with a baroclinic semi-Lagrangian spectral model. Mon. Wea. Rev.,122, 2391–2399.

  • Searcy, C., K. Dean, and W. Stringer, 1998: PUFF: A high-resolution volcanic ash tracking model. J. Volcanol. Geotherm. Res.,80, 1–16.

  • Self, S., and G. P. L. Walker, 1991: Ash clouds: Characteristics of eruption columns. U.S. Geol. Surv. Bull.,2047, 81–86.

  • Sparks, R. S. J., M. I. Bursik, S. N. Carey, A. W. Woods, and J. S. Gilbert, 1991: The controls of eruption-column dynamics on the injection and mass loading of ash into the atmosphere. U.S. Geol. Surv. Bull.,2047, 81–86.

  • Suzuki, T., 1983: A theoretical model for dispersion of tephra. Arc Volcanism: Physics and Tectonics, D. Shimozuru and I. Yokoyama, Eds., Terra Publishing Co., 95–113.

  • Tremback, C. J., J. Powell, W. R. Cotton, and R. A. Pielke, 1987: The forward-in-time upstream advection scheme. Mon. Wea. Rev.,115, 540–555.

  • Tripoli, G. J., and W. R. Cotton, 1980: A numerical investigation of several factors contributing to the observed variable intensity of deep convection over south Florida. J. Appl. Meteor.,19, 1037–1063.

  • Tripoli, G. J., and W. R. Cotton, 1982: The Colorado State University three-dimensional cloud/mesoscale model—1982. Part I: General theoretical framework and sensitivity experiments. J. Rech. Atmos.,16, 185–220.

  • Vandemeulebrouck, J., A. W. Hurst, and N. Poussielgue, 1994: Implications for the thermal regime of acoustic noise measurements in Crater Lake Mount Ruapehu, New Zealand. Bull. Volcanol.,56, 493–501.

  • Walko, R. L., and C. J. Tremback, 1995: HYPACT: The Hybrid Particle and Concentration Transport Model. User’s guide, 13 pp. [Available from ASTER Division, Mission Reasearch Corporation, P.O. Box 466, Fort Collins, CO 80522.].

  • Wilson, C. J. N., 1991: Ash-fall deposits from large-scale phreatomagmatic volcanism: Limitations of available eruption-column models. U.S. Geol. Surv. Bull.,2047, 93–99.

Fig. 1.
Fig. 1.

Isopachs (mm) for three tephra falls of the 1995 and 1996 Ruapehu event. [Figure adapted from Cronin et al. (1998)].

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 2.
Fig. 2.

Map showing the topography of the upper half of New Zealand (contour interval is 400 m) and the boundaries of the two inner RAMS grid (indicated by R2 and R3) and the boundary of the ASHFALL Model (indicated by ASH1). The other letters mark locations mentioned in the text as follows: R, Mount Ruapehu; A, Auckland; P, Paraparaumu; NP, New Plymouth; and G, Gisborne.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 3.
Fig. 3.

Terminal velocity as a function of particle size (μm) as computed by the HYPACT model.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 4.
Fig. 4.

National Oceanic and Atmospheric Administration visible satellite image for 1512–1516 NZST (=0312–0316 UTC) 17 Jun 1996. Notice the shadow cast by the ash cloud southeast of Lake Rotorua as indicated by the solid arrow. Note also the bright top of the eruption column as indicated by the dotted arrow.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 5.
Fig. 5.

RAMS model–derived skew T soundings of temperature, dewpoint temperature, and winds for a point near Mount Ruapehu for (a) 1200 NZST 17 Jun 1996, (b) 0000 NZST 12 Oct 1995, and (c) 1800 NZST 14 Oct 1995. The labels on the vertical axis are pressure (hPa), the labels on the horizontal axis are temperature (°C), and the isotherms slope upward from left to right.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 6.
Fig. 6.

RAMS/HYPACT-simulated distribution of settled ash particles at 2100 NZST 17 Jun 1996. The vertical source was assumed to be 3000 m thick, from 4000 to 7000 m in elevation.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 7.
Fig. 7.

Same as Fig. 6, but the vertical source was assumed to be 500 m thick, from 6750 to 7250 m in elevation.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 8.
Fig. 8.

RAMS/HYPACT-simulated (a) horizontal distribution of ash particles at 1516 NZST 17 Jun 1996 and (b) vertical distribution of ash particles at 1500 NZST 17 Jun 1996. The vertical source was assumed to be 500 m thick, from 6750 to 7250 m in elevation.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 9.
Fig. 9.

Distance that ash of various particle sizes (μm) was transported downwind of Mount Ruapehu as computed by the HYPACT model using high-resolution RAMS-simulated winds for 17 Jun 1997.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 10.
Fig. 10.

Isopach (mm) distribution for the 17 Jun eruption as simulated by the ASHFALL Model using fall velocity distribution 1 and the observed winds from the 1200 NZST 17 Jun rawinsonde launch. The location of Paraparaumu is marked by the P in Fig. 2.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 11.
Fig. 11.

Isopach (mm) distribution for the 17 Jun eruption as simulated by the ASHFALL Model using fall velocity distribution 1 and the high-resolution RAMS-simulated winds for the location marked R in Fig. 2.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 12.
Fig. 12.

RAMS/HYPACT-simulated distribution of settled ash particles at 0800 NZST 12 Oct 1995. The vertical source was assumed to be 3000 m thick, from 7000 to 10 000 m in elevation.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 13.
Fig. 13.

RAMS/HYPACT-simulated distribution of ash particles above 1500 m elevation at 2300 NZST 11 Oct, and 0100 and 0500 NZST 12 Oct 1995. The orientation of the plume rotated clockwise with time. The vertical source was assumed to be 3000 m thick, from 7000 to 10 000 m in elevation.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 14.
Fig. 14.

Isopach (mm) distribution for the 11 Oct 1995 eruption as simulated by the ASHFALL Model using fall velocity distribution 1 and the high-resolution RAMS-simulated winds for the location marked R in Fig. 2.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 15.
Fig. 15.

RAMS/HYPACT-simulated distribution of (a) settled ash and (b) airborne ash particles at 0000 NZST 15 Oct 1995. The vertical source was assumed to be 3000 m thick, from 7000 to 10 000 m in elevation.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Fig. 16.
Fig. 16.

Isopach (mm) distribution for the 14 Oct 1995 eruption as simulated by the ASHFALL Model using fall velocity distribution 1 and the high-resolution RAMS-simulated winds for the location marked R in Fig. 2.

Citation: Journal of Applied Meteorology 40, 1; 10.1175/1520-0450(2001)040<0056:FIVADF>2.0.CO;2

Table 1.

The assumed distribution of fall velocities (m s−1) for ash particles, and the equivalent particle diameters (using the Stokes formulation) used for the various ASHFALL simulations.

Table 1.
Table 2.

The final plume height of the eruption column, given the RAMS-simulated sounding for 1200 NZST 17 Jun 1996, for various combinations of initial plume radius (r0), initial upward velocity (u0), initial plume temperature (θ0), and condensation rate (τ). Also shown are the final plume heights for one combination of these parameters for the 0600 NZST 12 Oct and 1600 NZST 14 Oct 1995 model soundings.

Table 2.
Table 3.

The mean direction of movement of the ash plume centerline for various altitudes and eruption times of ash release for the three major ash-producing episodes.

Table 3.
Save