Characterizing lithological, weathering, and hydrothermal alteration influences on volcanic rock properties via spectroscopy and laboratory testing: a case study of Mount Ruapehu volcano, New Zealand

Abstract

The geomechanical characterization of volcanic material has important implications for geothermal and mineral exploration, engineering design, geophysical signals of volcano unrest, and models of instability and mass flows. Chemical weathering and hydrothermal systems can alter the host rock, leading to changes in mechanical behavior and failure mode. Here, we compare the physical and mechanical properties of lava, autoclastic breccia, and pyroclastic (scoria) rocks from Mount Ruapehu volcano (Ruapehu) in New Zealand to mineralogical composition determined via infrared spectroscopy and scanning electron microscopy (SEM) with energy-dispersive X-ray spectroscopy (EDS). We use correlation matrices, principal component analysis, and parametric analysis to determine which parameters can be used to predict physical and mechanical properties and form the basis for transfer functions. Laboratory-based spectroscopy shows that the samples contain absorption features indicative of Al- and Mg-rich hydrous phyllosilicates (e.g., kaolinite, halloysite, montmorillonite), Fe- oxides (e.g., goethite), and sulfates attributed to surface weathering, supergene, and steam-heated alteration. We find that porosity and primary lithology are the predominant control on physical and mechanical properties, followed by the pervasiveness of weathering/alteration, and then mineralogical composition. Several properties, such as porosity, uniaxial compressive strength, P-wave seismic velocity, density, and Young’s modulus, show strong correlations with other properties, indicating the potential for transfer functions between these properties. Hydrothermally altered rocks near the vent complex (up to ~ 400 m depth beneath the crater lake) with high-intensity hydrothermal alteration do not follow typical physical and mechanical property trends due to high clay content, low permeability, and low strength. The presence of these rocks within the edifice at Ruapehu implies local barriers to fluid flow and subsequent pore pressure variations. Additionally, they may have less than half the strength than would be dictated by typical porosity-strength trends for surface rocks, increasing the likelihood of structural failure. Trends in the pervasiveness of weathering with physical and mechanical properties, along with shifts in the position of spectral absorption peaks as hydrothermal/weathering alteration increases, suggest that it may be possible to extrapolate properties from imaging spectroscopy.

Introduction

Chemical weathering and hydrothermal alteration influence the physical and mechanical properties of volcanic rock, with implications for geothermal and epithermal mineral resources (Heap et al. 2019a), the use of volcanic material in the construction industry (Yildiz et al. 2010), failure modes and the evolution of rock physical properties (Heap et al. 2015c), and volcano structural stability (Watters et al. 2000). In active volcanic environments, rock properties are critical for modeling instability or mass flows, and interpreting geophysical signals of volcano unrest (e.g., volcano-seismic or geodetic; Reid et al. 2001; Mordensky et al. 2019a; Heap et al. 2020; Hickey et al. 2020). Inherent heterogeneities in volcanic primary material create physically and mechanically varied structures (Mordensky et al. 2018; Saubin et al. 2019). These heterogeneities are exacerbated by the circulation of hydrothermal fluids and chemical weathering under widely varying temperature (surface temperature to > 400 °C), chemical, temporal, and spatial conditions. These processes result in mineral oxidation, dissolution, replacement, and/or precipitation that variably alter the physical and mechanical properties of the volcanic host rock (e.g., Reid et al. 2001; Dobson et al. 2003; Ball et al. 2013; Siratovich et al. 2014; Wyering et al. 2014; Mordensky et al. 2018, 2019a, 2019b; Farquharson et al. 2019; Heap et al. 2019a, 2019b; Kennedy et al. 2020). For example, weathering and alteration can decrease or increase porosity and permeability of volcanic rock depending on the nature of the host rock, the fluid type and composition, temperature, and the duration and manner of rock-fluid interaction (Frolova et al. 2015; Farquharson et al. 2019; Mordensky et al. 2019b; Villeneuve et al. 2020). Changes in these properties in turn influence fluid flow, strength, and the deformation response of rock masses (Mordensky et al. 2022). A considerable presence of clays due to hydrothermal alteration has been attributed to large-scale volcano collapses (López and Williams 1993; Crowley and Zimbelman 1997), supported by numerical simulations of collapse scenarios (Reid et al. 2001; Ball et al. 2018).

Here, we explore the influence of weathering and hydrothermal alteration on the physical and mechanical properties of lava, autoclastic breccia, and pyroclastic (scoria) rocks from Mount Ruapehu (Ruapehu), an active andesitic stratovolcano in New Zealand. Ruapehu is a glaciated volcano with both active and relict hydrothermal systems with variable surface exposure (Kereszturi et al. 2020; Douglas et al. 2022), resulting in a variety of fresh, weathered, and hydrothermally altered material (Townsend et al. 2017). Rock alteration mineralogy is determined using a combination of infrared spectroscopy and scanning electron microscopy (SEM) with energy-dispersive X-ray spectroscopy (EDS). Infrared spectroscopy collects information from hundreds of narrow and contiguous spectral bands in the visible (400–700 nm), near-infrared (NIR; 700–1000 nm) and shortwave infrared (SWIR; 1000–2500 nm) wavelengths, capable of identifying chemical characteristics of materials using indicator minerals with characteristic absorption features (van der Meer 2018). Many indicator minerals are common weathering, oxidation, and hydrothermal alteration products in volcanic environments. For example, iron is highlighted in the NIR region, useful for detecting minerals such as goethite, hematite, and jarosite (Zimbelman et al. 2005). Al-, Fe-, Mg-, -OH, -SO₄, -CO₃ and -H₂O-bearing minerals can be detected in the SWIR region, identifying minerals such as alunite, gypsum, anhydrite, topaz, muscovite, biotite, epidote, calcite, dolomite, and clay minerals (Hunt and Ashley 1979; Swayze et al. 2014; Neal et al. 2018). Infrared spectroscopy can also be sensitive to other mechanically relevant rock microstructure properties such as grain size and crystallinity (Clarke 1999; Ruitenbeek et al. 2019; Okada et al. 2020).

The aims of this paper are to: (1) produce a spectroscopic and petrologic description of the fresh, weathered, and hydrothermally altered rock of Ruapehu volcano, (2) physically and mechanically characterize these rocks, and (3) examine the relationships between physical and mechanical properties and the type and extent of weathering and hydrothermal alteration. Mineralogical-physical–mechanical relationships will help to develop geotechnical models of volcanic and hydrothermal systems and guide numerical simulations of volcanic hazards.

Materials and methods

Study area and materials

Ruapehu is a complex andesitic composite volcano located in the Taupo Volcanic Zone of the central North Island of New Zealand (Fig. 1). The volcano is frequently active, with several historic small-to-moderate explosive eruptions and lahar events (Scott 2013). The oldest geologic rocks are dated at 250 ky and were exposed to glaciation up until 10 ky (Hackett 1985), and the youngest fresh magmatic rocks were erupted in 1996 (Nakagawa et al. 1999). The stratigraphic framework of Ruapehu comprises four formations, based on geochronology, geochemistry and stratigraphic relationships, pinpointing distinct spatial–temporal stages of volcano evolution (Fig. 1): Te Herenga (200 to 150 ka), Wahianoa (166–80 ka), Mangawhero (50–15 ka) and Whakapapa (< 15 ka) formations (Hackett and Houghton 1989; Price et al. 2012; Conway et al. 2016; Townsend et al. 2017; Fig. 1c). The volcano has an active hydrothermal system underneath the summit vent that hosts a crater lake (Christenson and Wood 1993; Christenson et al. 2010). Mineral assemblages of ejecta indicate local temperatures greater than 750 °C, reflecting vapor phase or contact metamorphism within the vent associated with high-level magma emplacement (Christenson and Wood 1993). A physical model of the vent system from Christenson and Wood (1993) describes a deeper central vent complex with a northern-inclined vent offshoot originating around 350 m below the Crater Lake. Fossil hydrothermal systems are also variably exposed at the surface around the summit plateau (see areas of argillic alteration outside of the currently active Crater Lake area, Fig. 1b; Kereszturi et al. 2020; Douglas et al. 2022). This varied geological, glacial, and hydrothermal history results in a wide variety of fresh, weathered, and altered material.

Fig. 1

Location and variability of samples with respect to spectral properties and geological formations. (a) Topographic map. (b) Simplified geological map, with the major geological formations (after Townsend et al. 2017). (c) Alteration map of the summit plateau based on airborne hyperspectral imagery from Kereszturi et al. (2020). Note not all sample locations are included in this zoomed view of the summit plateau

Sampling locations were partially directed by aerial hyperspectral imagery collected using a push-broom, VNIR-SWIR AISAFenix imaging system, flown in March of 2018 (Kereszturi et al. 2020; Miller et al. 2020; Fig. 1b). A total of 23 blocks of lava, autoclastic breccia, and pyroclastic (scoria) rocks varying from fresh to pervasively altered were selected to represent the variability of the physical properties of the material forming the volcanic edifice (Fig. 1, Fig. 2, Table 1). The blocks had a volume of c. 8000 cm³ (typical edge length of 20 cm) and were large enough to obtain all the test specimens of a given rock type from the same piece. Each block was inspected for macroscopic defects so that it would provide test specimens free from fractures, seams, partings, or joints. Block sample names follow the same nomenclature as (Kereszturi et al. 2020). All blocks are considered surface samples excluding RH50; RH50 is a ballistic block from the 1995–1996 eruption, composed of moderately to highly altered andesites with pores and vesicles rarely filled with elemental sulfur and associated anhydrite, pyrite, and natroalunite (Christenson et al. 2010). During this eruption, the introduction of gas/fluid at depth promoted catastrophic failure of a partial mineralogic seal developed at the top of the hydrothermal systems in the northward-inclined vent beneath Crater Lake (Kilgour et al. 2010). Estimates of penetration depth into the vent complex indicate depths between 275 and 410 m, and the formerly molten elemental sulfur indicates a minimum pre-eruption temperature in the source area of 119 °C (Christenson et al. 2010). Thus, the sample originated within the current vent complex (Christenson and Wood 1993), and is considered a subsurface sample. Pyroclastic sample RH15 is a block of scoria located in an undifferentiated tephra fall deposit. The term ‘sample’ refers to the 23 rock blocks, while cores or pieces from each block are referred to as ‘specimens’ throughout.

Fig. 2

Photographs of the samples and porosity as determined by helium pycnometer

Table 1 Location, geologic formation, and geologic member of samples used in this study. A description of the formations and members can be found in (Townsend et al. 2017). Undiff = undifferentiated

Laboratory spectroscopy and Scanning Electron Microscopy (SEM)

Each of the 23 blocks listed in Table 1 was measured using an ASD FieldSpec® 4 Hi-Res NG spectroradiometer, which recorded 2151 spectral bands between 350 and 2500 nm. This instrument has a full-width-half-maximum (FWHM) of 3 nm at 700 nm and 8 nm at 1400–2100 nm regions. The spectral bandwidth is 1.4 nm between 350–1000 nm, as well as 1.1 nm between 1001 and 2500 nm. Spectral data were captured using a High Brightness contact probe with a halogen light illumination source. The contact probe has a measurement spot size of 10 mm. Each measurement was calibrated using a white Spectralon® Diffuse Reflectance Standard. Both freshly chipped surfaces (sample interior) and the exterior of the sample were measured for a total of 3–5 spectral readings per block. Phenocrysts are typically 1 mm or smaller (Fig. 3); therefore, the contact probe size, being 10 times larger than the average phenocryst size, was sufficient to capture the spatial heterogeneity in each block. See Danner et al. (2015) for FieldSpec schematic and spectral sampling details.

Fig. 3

A (a) Continuum-removed spectral reflectance curves of typical alteration minerals in volcanic environments from non-altered (RH17) to pervasively altered (RH50)- see descriptions in Table 3. Note the spectral shift from 550 to 480 nm where hematite is being replaced by goethite as a surface weathering product. The curves are offset for clarity by 0.1 percent. Each spectral curve has a maximum reflectance value of ‘1’, and the following minimum values: RH50 = 0.74, RH38 = 0.80, RH28 = 0.51, RH22 = 0.79, RH6 = 0.67, RH17 = 0.77. (b-e) Optical microscope and SEM images demonstrating textural and mineralogical changes through hydrothermal and weathering alterations. Labeled mineral phases: v – void/vugg, m – titanomagnetite, plg – plagioclase, opx – orthopyroxene, cpx – clinopyroxene, sm – smectites, kao – kaolinite, qtz – quartz, geo – goethite, py – pyrite, al – alunite, hem – hematite, s – sulfur

Inherent variation in detector sensitivity and variations in temperature conditions can cause spectral drift at two wavelength locations where detector arrays meet (e.g., VNIR and SWIR). These drifts were corrected by applying a splice correction function using the ASD ViewSpec Pro software (Danner et al. 2015). The splice correction function considers the average of tangents at either side of a break point to determine the new points through which the line passes without drift (see Danner et al. (2015) for schematic and details). The splice-corrected spectral readings for each block were then averaged. The spectral data were then normalized using a convex hull continuum removal method. The continuum is a convex “hull” of straight-line segments fitted over the top of the spectrum that connect local spectral maxima. The continuum is removed by dividing the original reflectance value by the corresponding values of the continuum line (convex hull) at a given wavelength. Removing this continuum standardizes isolated absorption features for comparison between samples (Kokaly and Clark 1999; Huang et al. 2004).

The sample mineralogy was further constrained using a ThermoFisher Scientific™ FEI Quanta 200 Environmental Scanning Electron Microscope (SEM) operated in Back-Scattered Electron (BSE) mode under accelerating voltage of 20 kV, with a working distance of 10 mm, at Massey University’s Manawatu Imaging Centre. Minerals were recognized based on their textural and crystal habits as well as using Energy-dispersive X-ray Spectroscopy (EDS) (Shindo and Oikawa 2002; Severin 2004).

Laboratory physical and mechanical measurements

The 23 blocks were cored to create several specimens per block (5 to 8) with a diameter of 20 mm, and then ground to a length:diameter ratio of 2–2.15:1 (see Supplementary Table 1 for details of each specimen). For each specimen, we measured porosity, permeability, and compressional (Vp) and shear wave (Vs) velocity. Dynamic Young’s modulus and dynamic Poisson’s ratio were calculated from Vp and Vs results. Porosity, permeability, Vp, Vs, dynamic Young’s modulus, and dynamic Poisson’s ratio values were then averaged for each block. Magnetic susceptibility was measured on the natural face and a cut face of each block using a magnetic susceptibility meter and averaged to produce one value of magnetic susceptibility per block. Two to three specimens from each block were then used for uniaxial compressive strength (UCS) experiments, from which static Young’s modulus values were calculated. UCS and static Young’s modulus were also averaged to produce a single value for each block. Another three specimens from each block were used for triaxial compressive strength testing. Triaxial results were used to calculate friction angle, cohesion, and the Hoek–Brown material constant for intact rock (m_i) for each block. All specimens were oven-dried at 60 °C for a minimum of 48 h and then allowed to cool to room temperature prior to testing.

All physical and mechanical property testing was conducted at the University of Canterbury rock mechanics laboratory. The physical and mechanical properties measured and examined in this study are commonly used to develop models of fluid flow (e.g., porosity, permeability; Day 1996; Gonnermann and Manga 2007; Kennedy et al. 2020), slope failure (e.g., UCS, friction angle, cohesion, material constant for intact rock m_i; Apuani et al. 2005; Schaefer et al. 2013; Heap et al. 2021; Wallace et al. 2021), subsurface imaging, and monitoring deformation and geophysical phenomena such as seismicity (e.g., Young’s modulus, Poisson’s ratio, Vp, Vs, density, magnetic susceptibility; Lu et al. 2005; Mordensky et al. 2019a; Heap et al. 2020).

Pearson’s correlation coefficient (PCC or Pearson’s r) and Spearman’s rank correlation coefficient (SCC or Spearman’s ρ) were used to correlate one physical or mechanical property (X) to another physical or mechanical property (Y). Pearson’s correlation coefficient is the measure of the linear correlation between two variables as:

$$PCC {\left(r\right)}_{X,Y}=\frac{cov(X, Y)}{{\sigma }_{X}{\sigma }_{Y}}$$

(1)

where cov is the covariance, σ_X is the standard deviation of X and σ_Y is the standard deviation of Y. Spearman’s rank correlation coefficient is the measure of the monotonic correlation between two variables. For a sample size of n, the n raw scores X_i and Y_i are converted to ranks R(X_i) and R(Y_i), and SCC is computed as:

$$SCC {\left(\rho \right)}_{R\left(X\right), R(Y)}=\frac{cov(R(X),R(Y))}{{\sigma }_{R(X)}{\sigma }_{R(Y)}}$$

(2)

where the cov is the covariance of the rank variable, and the σ_R(X) and σ_R(Y) are the standard deviation of the two rank variables being compared. The correlation coefficient values (r or ρ) range from -1 to 1, with ‘-1’ indicating a total negative correlation between variables, ‘0’ indicating no relationship, and ‘1’ indicating a total positive relationship. The strength of the relationship has different interpretations among scientific research areas (see Akoglu 2018); our interpretations of the Pearson’s and Spearman’s correlation coefficients are as follows: ( −)0.10 to ( −)0.39 is a “weak” relationship, ( −)0.40 to ( −)0.59 is a “moderate” relationship, ( −)0.60 to ( −)0.79 is a “strong” relationship, and ( −)0.80 to ( −)1.0 is a “very strong” relationship.

In addition to the correlation coefficients, we conducted a principal component analysis (PCA), a multivariate technique that reduces the dimensionality of large datasets via linear transformation, which allows for feature extraction and data visualization (Jolliffe 2005). Herein, PCA is used to describe which physical and mechanical parameters have similar contributions to variance in the dataset..

Porosity, density, and permeability

Oven-dried specimens were used for measuring dry mass, dry density, unit weight, porosity, and permeability. Porosity was measured using a Micromeritics AccuPyc™ II 1340 helium pycnometer, which measures the specimen’s solid volume (V_s). This is subtracted from the total volume (V_t) to determine pore volume (V_p) and connected porosity (φ), as follows: V_t-V_s = V_p, and φ = V_p/V_t. Specimen density was calculated by dividing the mass of each specimen by its total volume (V_t).

For relatively high permeability specimens (> 10¹⁷ m²), permeability was measured at room temperature using a steady-state Vinci Technologies benchtop permeameter interfaced with a Bronkhorst El-Flow® volumetric flowmeter. The steady state permeameter records the volumetric flow rate through the specimen core, driven by the pressure differential of nitrogen as the pore fluid upstream, and ambient atmospheric pressure downstream. Specimens were placed in a rubber sleeve in a Hoek cell and radially confined at 1 MPa prior to applying nitrogen gas at a steady pressure to one end of the sample. The radial confinement assured that the nitrogen gas flowed through the specimen and not between the sample edge and the membrane. See Hill (2020) for steady state permeameter setup details. Flow rate measurements were collected at several pressure gradients, allowing permeability to be calculated using Darcy’s Law:

$$-\frac{dp}{L}=\frac{(\mu Q)}{A{k}_{D}}$$

(3)

where p is the pressure, L is the length of the specimen, A is the cross-sectional area, k_d is the gas permeability, μ is the dynamic viscosity coefficient of the pore fluid, and Q is the flow rate.

For lower permeability specimens (< 10^–17 m²), permeability was measured using a Core Laboratories PDP-200 pulse decay permeameter at 30 °C. Specimens were placed in a core holder and a uniform confining pressure of ~ 2 MPa was manually applied with a hydraulic pump to assure that the test fluid moved through the specimen and not between the specimen and the core holder. The system uses nitrogen gas to saturate the test specimen and then the downstream gas valve is opened, allowing for a pressure differential to develop across the specimen. The system then measures the pressure differential decay across a specimen at regularly timed intervals. Permeability was calculated through a function provided in Brace et al. (1968) as follows:

$${k}_{gas}=\left(\frac{2\eta L}{A}\right)\left(\frac{{V}_{up}}{{p}_{up}^{2}-{p}_{down}^{2}}\right)\left(\frac{\Delta {p}_{up}}{\Delta t}\right)$$

(4)

where k_gas is gas permeability, ŋ is the viscosity of the pore fluid (nitrogen), ∆t is change in time, L is the length of the specimen, A is the cross-sectional area of the specimen, V_up is the volume of gas in the upstream reservoir and piping, p_up is the pressure of the gas in the upstream reservoir and piping, p_down is the pressure of gas in the downstream reservoir and piping, and ∆p_up is the change in p_up during the elapsed time. See Cant et al. (2018) for pulse-decay permeameter setup details.

Magnetic susceptibility

Magnetic susceptibility was measured using a Terraplus KT-10 v2 magnetic susceptibility meter on the natural face and a cut face of each block. The KT-10 m was held in direct contact with either the natural or cut surface of the block in accordance with the manufacturer (see Terraplus User’s Guide, v. 2.1), measured three times, and averaged to produce one value per block. Measurements were not made on individual specimens due to the 65-mm coil diameter in the KT-10 m.

Elastic wave velocities and dynamic elastic moduli

Dry compressional (Vp) and shear wave velocities (Vs) were collected using a GCTS CATS interfaced with transducer-mounted piezoelectric quartz crystals operating at a 900 kHz resonance frequency and a 20 MHz pulse sampling rate, with a minimum of 100 waveforms per compression wave type per specimen. Ultrasonic gel and a constant 312 Pa stress were applied to ensure sufficient contact between the specimens and platens to produce consistent waveforms. Dynamic Young’s modulus and dynamic Poisson’s ratio were calculated as follows (Guéguen and Palciauskas 1994)

$${\nu }_{d}=\frac{{V}_{P}^{2}-{V}_{S}^{2}}{2{(V}_{p}^{2}-{V}_{s}^{2})}$$

(5)

$${E}_{d}=\frac{\rho {V}_{s}^{2}\left(3{V}_{p}^{2}-4{V}_{s}^{2}\right)}{\left({V}_{p}^{2}-{V}_{s}^{2}\right)}$$

(6)

where ${\nu }_{d}$ is the Poisson's Ratio, V_p is compressional wave velocity, V_s is shear wave velocity, E_d is the Young’s Modulus, and ρ is dry bulk density.

Uniaxial compressive strength and static Young’s modulus

Dry uniaxial compressive strength (UCS) measurements were conducted on a Technotest 3000 kN servo-controlled loading frame at room temperature. Specimens were tested at a constant strain rate of 1.0 × 10^–5 s⁻¹, like other studies of volcanic material (Siratovich et al. 2014; Heap et al. 2015b; Schaefer et al. 2015; Mordensky et al. 2018). Axial strain was measured using a linear-variable displacement transducer extensometer, and static Young’s modulus was determined using the average modulus of the linear portion of the stress–strain curve following ASTM standards (ASTM D7012-07).

Triaxial compressive strength, friction angle, cohesion, and intact rock constant m_i

Three specimens from each block were tested in conventional triaxial conditions (${\sigma }_{1}^{^{\prime}}>{\sigma }_{2}^{^{\prime}}={\sigma }_{3}^{^{\prime}}$, in which ${\sigma }_{1}^{^{\prime}}$ is the major effective principal stress, ${\sigma }_{2}^{^{\prime}}$ is the intermediate effective stress, and ${\sigma }_{3}^{^{\prime}}$ is the minor effective principal stress) over a limited range of confining stresses (${\sigma }_{3}^{^{\prime}}$), with no confining stress larger than 50% of the average UCS of the block (as determined using by calculating the average UCS of specimens tested from the same block) to ensure that failure remains in the brittle regime (Labuz and Zang 2012; Hoek and Brown, 1980; Hoek and Brown 1997). Dry triaxial compressive strength measurements were conducted using a Hoek cell and deformed with a Technotest 3000 kN servo-controlled loading frame at a constant strain rate of 1.0 × 10^–5 s⁻¹ at room temperature. Common confining stresses of 2, 3, 5, 10, 20, 22, or 25 MPa were used, when possible, to allow for comparison between samples. Results from the triaxial laboratory experiments were analyzed with the Rocscience RSData software to calculate material constant m_i, friction angle and cohesion for each block. The material constant m_i was calculated using the Hoek–Brown failure criterion for intact rock with the modified cuckoo curve fitting method and basic error summation method (see Supplementary Table 2 for RocData processing results). Friction angle and cohesion were calculated using the linear curve fitting method and vertical error summation method (see Supplementary Table 3 for RocData processing results).

Results

Blocks were categorized by volcanic lithology, which distinguishes rocks by physical characteristics resulting from depositional or emplacement processes. Herein, we use the following terms: ‘lavas’ are former molten rocks expelled at the Earth’s surface by volcanic processes; ‘autoclastic breccias’ are rocks formed during effusive volcanism when a lava flow cools and fragments in contact with air, and the fragments are deposited under continued lava flowage; ‘pyroclastic rocks’ are rocks containing fragments formed from volcanic plumes, jets, or pyroclastic density currents as particles first come to rest via suspension settling, traction, and/or en masse freezing (Fisher 1961; White and Houghton 2006). In total, we present the physical and mechanical characteristics of 17 lavas (one being ejected block RH50 as described in the Materials and Methods section), 5 autoclastic breccias, and one pyroclastic rock, which is a scoria. While the pyroclastic sample is included in the following presentation of the laboratory results, we caution the reader on over-interpreting patterns related to pyroclastic rock, since there is only one sample in this lithologic category.

Alteration mineralogy and classification

The laboratory-based spectroscopy results show that the blocks contain distinct absorption features around 1410–1450 nm, 1920 nm, and within the broader spectrum between 2180–2400 nm (Fig. 3a), indicating the presence of phyllosilicates (clays) including kaolinite, halloysite, smectites (mostly montmorillonite and nontronite), and minor illite (see Douglas et al. 2022, and references therein). The samples also occasionally contain minor absorption features around 2180 and 2265 nm, which are due to the presence of alunite and jarosite, respectively (Fig. 3a). More commonly, the samples have absorption features in the visible region, notably around 390, 425, 485, and 540 nm, which are consistent with minerals such as sulfur, goethite, and hematite (Fig. 3a). Based on SEM–EDS analysis, some samples have disseminated pyrite, occurring together with silica polymorphs (e.g., RH28, RH38, Fig. 3d, e). These minerals do not have any strong characteristic absorption features in the visible and shortwave-infrared regions. Alteration minerals are often limited to vugg and fracture infills and occur as replacement products of the rock’s matrix, and partial to full replacement of the original mineralogy (Fig. 3d, e). A detailed characterization of sample alteration mineralogy, including a conceptual model of alteration history at Ruapehu, is described in detail in Kereszturi et al. (2020).

Following the weathering/alteration assemblages, defined by indicator minerals (minerals specific to certain weathering, oxidation, and hydrothermal alteration processes), their occurrence, and their origin (Rye 2005; Simmons et al. 2005) samples were categorized into four mineral assemblages (Table 2): (1) Fresh, consisting of primary mineralogy such as pyroxene, plagioclase, and titanomagnetite; (2) Surface weathering, consisting of primary and chemically weathered and oxidized minerals including Fe-oxides (hematite) and minor smectites typically occurring in a weathered rim; (3) Supergene, argillic alteration that develops at < 40 °C through long-term weathering and oxidation of ferrous and minor sulfide-bearing rocks in atmospheric conditions, which includes phyllosilicates, jarosite, and Fe-oxides; (4) Steam-heated, intermediate and advanced argillic alteration that develops at < 120 °C near the water table and in the shallowest epithermal environment, including, quartz, pyrite, Fe oxides (goethite), phyllosilicates, sulfur, and occasional jarosite and alunite (Table 2). The latter two categories often show mineral overprinting, when rocks subjected to hypogene conditions are later subjected to weathering and erosion, resulting in the replacement of some meta-stable hydrothermal minerals (e.g., pyrite to jarosite; schwertmannite and goethite/hematite; anhydrite to gypsum) in atmospheric pressure and temperature conditions.

Table 2 Weathering/alteration category and associated mineral assemblages of collected sample blocks

In addition to various mineral assemblages, samples also showed variations in alteration pervasiveness. Thus, samples were also categorized into four groups based on the extent of weathering/alteration, including “none” (fresh material), “minor” (typically only the exterior/crust of the sample is weathered or altered), “moderate” (weathering or alteration is present throughout the groundmass), and “pervasive” (weathering or alteration is strongly present in groundmass and crystals) (Table 3). Surface weathering samples have varying percentages of total weathered material, with autoclastic breccias (e.g., samples RH4, RH6) and the tephra (RH15) being more pervasively weathered than lavas (e.g., samples RH2b, RH2, RH5, RH8, RH10, RH16) due to pre-existing fractures, smaller grain size, and more void space.

Table 3 Descriptions and average physical and mechanical laboratory results for each sample block

Physical and mechanical characterization

Block averaged properties are shown in Table 3, and properties for each specimen (individual cores) are available in Supplementary Table 1. Individual parameters are plotted against either porosity (Figs. 4 and 5) or Vp (Fig. 6), which systematically have the highest correlation factors to other physical and mechanical properties (Fig. 7).

Fig. 4

Porosity vs. uniaxial compressive strength (UCS) of all tested specimens, split into (a) lithology, (b) weathering/alteration mineralogy, and (c) weathering/alteration mineralogy. Porosity-UCS trend outliers RH50 and RH52 are discussed in the text

Fig. 5

Relations of various averaged block properties vs. averaged block porosity, split into columns according to lithology, weathering/alteration type, and weathering/alteration pervasiveness. (a-c) Porosity vs. permeability. (d-f) Porosity vs. magnetic susceptibility. (g-i) Porosity vs. static Young’s modulus. (j-l) Porosity vs. Generalized Hoek–Brown failure criterion intact rock constant m_i. (m–o) Porosity vs. Vs

Fig. 6

Relations of various averaged block properties vs. averaged block P-wave velocity (Vp), split into columns according to lithology, weathering/alteration type, and weathering/alteration pervasiveness. (a-c) Vp vs. porosity. (d-f) Vp vs. static Young’s modulus. (g-i) Vp vs. cohesion. Outlier RH50 in pervasiveness of weathering/alteration graphs is discussed in the text

Fig. 7

Correlation matrices between average block parameters, showing (a) linear (Pearson) and (b) monotonic (Spearman rank) relationships between variables. Legend indicates strength of correlation coefficient (ranging from black/value of 1: high correlation to white/value of 0: weak correlation). Sample RH50, which originated from the upper conduit and thus is considered a subsurface sample, is not included in these matrices

Porosity, permeability, density, and magnetic susceptibility results are as follows:

• The average connected porosity of all specimens ranges from 1.6-48.3%, highlighting the heterogeneity of the sample suite (Figure 2, Figure 4, Table 3).

• Porosity is generally higher in autoclastic breccias and pyroclastic rocks (c. 22-48 %) than lavas (c. 2-30 %) (Figure 4a).

• Porosity is highest in rocks with surface weathering (average = 18%) (Figure 4b) and pervasive alteration (average = 28%) (Figure 4c).

• The standard deviation of porosity between specimens is below 2%, except for RH6 (SD = 3%) and RH19 (SD = 3.2%).

• Specimen dry mass density generally increases with decreasing values of porosity, ranging from 1.37 to 2.67 g/cm³.

• Permeability exhibits 6 orders of magnitude difference between the lowest (4x10^-18 m²) and highest (3x10^-12 m²) permeable specimens (Figure 5a-c).

• Lavas typically have lower permeability (<1x10^-14 m²) than autoclastic breccias and pyroclastic rocks (>1x10^-14 m²) (Figure 5a). Despite this trend, the permeability of rocks with similar porosities can vary by up to two orders of magnitude.

• Magnetic susceptibility ranges from 0 to 0.04 SI, with a weak relationship with porosity (Figure 5d-f).

• Pervasively altered and steam-heated blocks have lower magnetic susceptibility than less altered or unaltered blocks, and lavas have higher magnetic susceptibility than autoclastic breccias and pyroclastic rocks (Figure 5d-f).

Dry Vp and Vs are inversely related to porosity, with Vp and Vs velocities typically being higher in lavas than autoclastic breccias and in the pyroclastic rock (Fig. 5m-o, Fig. 6, Table 3); all lava Vp are > 3600 m/s and Vs are > 1300 m/s, while autoclastic breccias have Vp < 3400 m/s and Vs < 1400 m/s, apart from RH42b with a Vs of 1661 m/s. The relationship of Vp and Vs with porosity has been previously established (e.g., Heap et al. 2014a, b; Wyering et al. 2014; Mordensky et al. 2018). Additionally, Vp and Vs are lowest in pervasively altered samples, but do not systematically vary with alteration pervasiveness or alteration type (Fig. 6b-c), as has been shown by previous studies (Wyering et al. 2014). Similar patterns exist for dynamic Poisson’s ratio and dynamic Young’s modulus, which range between 0.23–0.46 and 7.1–32.1 GPa, respectively. Static Young’s modulus shows a wider range than dynamic Young’s modulus, varying between 3.7–42.3 GPa. Again, static Young’s modulus is significantly higher in lavas (average = 23 GPa) than in autoclastic breccias and the pyroclastic rock (average = 7.5 GPa), is closely dependent on porosity (Fig. 5g-i), shows weak systematic decreasing trends with increasing alteration type and pervasiveness, and has a strong relationship to Vp (Fig. 6d-f). Dynamic Young’s modulus values tend to be higher than static Young’s modulus, although this is reversed for stiffer material (values > 25 GPa) (Fig. 8).

Fig. 8

Static vs. dynamic Young’s modulus values. Dynamic Young’s modulus values tend to be higher than static Young’s modulus, although more stiff material (values > 25 GPa) tends to have higher static Young’s modulus values

Uniaxial compressive strength (UCS) correlates with porosity (Fig. 7), with UCS decreasing exponentially with increasing porosity (Fig. 4). In lower-porosity samples (< 10%), UCS can vary by over 200 MPa. Variations in UCS for samples with > 10% porosity are much smaller (~ 50 MPa max). UCS at low porosity (< 20%) decreases with an increase in the pervasiveness of weathering or hydrothermal alteration, and fresh material is stronger than surface weathering and supergene alteration, all of which are stronger than steam-heated material. Again, there can be large ranges of UCS values within these alteration categories, and the trends described do not apply to material >  = 20% porosity.

Triaxial experiment results are shown in Fig. 9 and summarized in Table 4. These experiments show that sample strength increases with increasing confining pressure (Table 4, Fig. 9e-f). Stress–strain curves show that fresh and altered lava samples exhibit brittle mechanical behavior (Fig. 9a) at low confinement, where a peak stress (differential stress) is reached followed by strain softening (as described in Wong and Baud 2012). While fresh breccias also display brittle mechanical behavior at low confinement, altered breccias display ductile behavior (Fig. 9b), characterized by an absence of strain softening and dilation, and where failure is a function of compactant pore collapse (as described in Heap et al. 2015a). Unaltered lavas tend to be strong, stiff and with low porosities, while altered lavas tend to be weak, ductile or have high porosities (Fig. 9a-b, e–f), and samples with similar porosities tend to have similar strength values at a given confining pressure (Fig. 9c-d). Altered samples tend to reach lower peak stress, have a less defined failure peak, or experience higher strains than fresh samples (Fig. 9a-b, e–f), although deviations from these patterns do occur.

Fig. 9

Stress–strain mechanical data from triaxial experiments. (a) Variably altered lava samples at a confining pressure (Pc) of 5 MPa (b) Variably altered autoclastic breccia samples also at Pc = 5 MPa. (c) Low porosity samples at Pc = 5 MPa (d) high porosity samples at Pc = 5 MPa. (e) Minor supergene and (f) moderate steam-heated lava sample behavior at three confining pressures

Table 4 Triaxial compressive strength results for varying confining pressures. Blank cells indicate that the confining pressure was not used for a given block

Cohesion covers a wide range of values from 1.8 to 52.4 MPa (Table 3), which has a positive correlation with UCS (Fig. 7) and a positive logarithmic relationship with Vp (Fig. 6g-i). Like UCS, cohesion is higher in lavas than autoclastic breccias and in the pyroclastic rock and decreases with increasing weathering and alteration pervasiveness. Friction angle ranges from 37 to 57° (Table 3) and has a positive and monotonic (in this case logarithmic) relationship with intact rock parameter m_i (Fig. 7). Intact rock parameter m_i ranges from 6 to 48 (Table 3) and has a weakly negative monotonic correlation with porosity (Fig. 7), but no clear relation to rock texture, alteration type or pervasiveness when plotted against porosity (Fig. 5j-l).

PCA results (Fig. 10) group the physical and mechanical parameters in terms of similar impacts on the variance of the dataset. The results are presented as a set of biplots for the first three principal components to encompass a significant proportion of the variance, as well as to identify principal components for which parameters correlate and do not correlate. The directions of the arrows represent the eigenvectors of the analysis. Parameters that correlate tend to have arrows pointing in the same or opposite directions. Parameters that do not correlate tend to have arrows pointing in orthogonal directions. The biplots show that, for three combinations of the first three principal components (representing 80% of the variance), parameter groupings remain quite consistent, although some of the trends are stronger for different combinations. The physical and mechanical property relations revealed in these graphs is discussed below.

Fig. 10

Biplots of eigenvectors resulting from principal component analysis. Colours are RGB coded according to the x, y and z components of the eigenvectors corresponding to the first three principal components. Similarly coloured parameters have eigenvectors with similar directions in the 3D space defined by the first three principal components

Discussion

Weathering and alteration mineralogy

The predominant weathered/altered minerals identified by infrared spectroscopy are Fe-bearing oxides and clay mineral associations typical of acid-sulfate alteration processes (e.g., kaolinite, smectites), producing argillic to advanced argillic alteration of the primary volcanic rocks (Heald et al. 1987; Simmons et al. 2005). The alteration minerals form through supergene to hypogene weathering and/or alteration of the host rocks that originally contain plagioclase, minor pyroxene, (titano-)magnetite, and rarely olivine, amphibole phenocrysts and microcrystals in the groundmass (Fig. 3). Phyllosilicates, sulfides, and sulfates are consistent with acidic and steam-heated environments, such as the currently active vent-hosted system underneath Ruapehu’s Crater Lake (Christenson and Wood 1993; Christenson 2000), which precipitates such alteration minerals via oxidation of H₂S and H₂SO₄-rich fluids by atmospheric oxygen at shallow depths.

Samples show a strong shift in the position of the Fe³⁺ iron absorption from 530 nm (e.g., hematite-dominated) to shorter wavelengths around 480 nm (e.g., goethite-dominated), as hydrothermal/weathering alteration increases (Fig. 3). This spectral shift is indicative of the pH preference of these indicator minerals (e.g., hematite-dominance in less-acidic and neutral pH conditions; Schwertmann and Murad 1983) and the presence of disseminated pyrite formed through acid-sulfate alteration, which may later undergo supergene oxidation, resulting in the formation of Fe-oxides (goethite), jarosite, gypsum, and smectite (Fernandez-Caliani et al. 2004; Rye 2005). Spectral peaks at 2160–2210 nm, 2315 nm, and 2390 nm for samples with increased clay content (e.g., kaolinite, smectites) are associated with longer exposures to weathering or hydrothermal fluids. This is consistent with the present-day geology of Ruapehu, where ancient hydrothermal systems can be partially covered by younger volcaniclastic and lava rocks (Kereszturi et al. 2020).

Physical and mechanical property relations

Physical and mechanical test results reveal several common relations between material properties (Tables 3 and 4, Figs. 4, 5, 6, 7, 8, 9 and 10). As shown in the PCA results (Fig. 10), measured UCS and calculated σ_ci always plot together, highlighting that they could be interchanged depending upon which is available. They tend to plot in the vicinity of the other strength parameters, cohesion, and elastic parameter static Young’s modulus. Friction angle and m_i also consistently plot close to each other because they represent the shape, and slope and curvature, respectively, of the failure criterion to which each belongs. Porosity and permeability also tend to coincide, whereas density and Vp consistently plot in the opposite direction, highlighting the inverse impact of porosity on density and Vp, and its positive impact on permeability. Dynamic Young’s modulus and Vs tend to plot in similar locations, which are often orthogonal to the other parameter directions, suggesting that there is low correlation to other parameters. Magnetic susceptibility does not have any consistent relationships with the other parameters; for some principal component combinations it plots near or parallel to some parameters (e.g., density and porosity in Fig. 10a), and for other combinations it plots away from all other parameters (e.g., Fig. 10c).

Similar to other studies, we find that porosity has a first-order effect on other physical and mechanical properties, including strength and stiffness, permeability, and rock failure mode (Al-Harthi et al. 1999; Pola et al. 2012; Heap et al. 2014a, 2014b, 2015b, 2019a; Schaefer et al. 2015; Siratovich et al. 2016; Cant et al. 2018; Coats et al. 2018; Mordensky et al. 2018). UCS and Young’s modulus decrease exponentially with increasing porosity, while permeability increases exponentially with porosity (Fig. 4, Fig. 5a-c, g-i). Low porosity samples also generally have higher values of Poisson’s Ratio, m_i, magnetic susceptibility, and Vp. Lavas typically have lower porosities (< 20%) than autoclastic breccias and the pyroclastic rock (> 20%), mostly due to emplacement mechanisms; the slow cooling of lava flows allows gas bubbles to escape and coalesce prior to solidification, resulting in lower porosity, while pyroclastic rocks freeze and fragment rapidly with the bubbles still distributed throughout the clast, resulting in higher porosity. Thus, lithology (i.e., lava versus autoclastic breccia) additionally correlates to material properties (Figs. 4, 5 and 6; as in Mordensky et al. 2018). However, it should be stressed that, within each of these categories, properties can vary significantly, as also highlighted by the variability in the direction of the density eigenvectors compared to the other parameters in Fig. 10. For example, in low porosity samples (lavas with porosity < 10%), UCS can vary by over 200 MPa. Even within 6–7% porosity, sample peak stress can vary by 100 MPa (Fig. 9c). Clearly, other parameters have more impact on strength at low porosity.

Vp, density, and Young’s modulus also show strong relationships with many material properties (Fig. 5g-i, Fig. 6, Fig. 7). Seismic velocities in volcanic rocks typically attenuate with the presence of glass, clay alteration, microcracks, and vesicles (Vanorio et al. 2002; Pola et al. 2012). Like other authors (Wyering et al. 2014), we find that the correlations between Vp and porosity, density, and permeability are strong. Correlations of Vs with other properties is weaker than Vp, likely because of errors related to the difficulty in picking the first arrival of the S-wave, leading to a large dispersion of Vs values (Vilhelm et al. 2016), which also affects the dynamic Young's modulus and dynamic Poisson's ratio, reducing potential for deriving transfer functions from these parameters. Static Young’s modulus correlates well with Vp, density, and strength (Fig. 7). Density, and dynamic Poisson’s ratio and Young’s modulus calculated from seismic velocities, show similar but weaker relations compared to static Young’s modulus (Fig. 7), likely arising from the dispersion of Vs. We find that magnetic susceptibility has a moderate to weak relationship with Young’s modulus (Fig. 7), and porosity and density (Fig. 5d-f, Fig. 7), although the relationship with Young’s modulus may be the result of porosity rather than a truly independent relationship. The correlation of magnetic susceptibility with porosity and density for a given rock type is well established (Henkel 1994); in our samples, it likely stems from magnetic mineral oxidation and alteration, as well as with lower porosity but also lower density, non-magnetic clay minerals replacing primary magnetic minerals (e.g., titanomagnetite). The relationship of magnetic susceptibility with other material properties is generally weak (Fig. 7, Fig. 10), although lavas tend to have higher magnetic susceptibility values than autoclastic breccias and the pyroclastic rock.

Friction angle and m_i correlate well to each other, as expected, but not to any other parameters (Fig. 7), except weakly with porosity (Fig. 5j-l; Villeneuve and Heap 2021). Interestingly, friction angle and m_i do not appear to correlate for combinations of PC1&PC2 nor PC1&PC3 but do correlate for the PC2&PC3 combination (Fig. 10), suggesting that generating a transfer function for friction angle and m_i is possible but not trivial. Cohesion correlates to UCS as expected, since they both describe the low-confinement strength components of the failure criterion to which each belongs, as well as Young’s modulus, density (Fig. 5g-i), Vp (Fig. 6d-f), and porosity (Fig. 7), as also shown in Villeneuve and Heap (2021). The PCA analysis shows, however, that the correlations between strength (UCS) and physical parameters are not consistent across the three principal components examined, suggesting complex relationships. This is the first research that attempts to link m_i to alteration and weathering of volcanic rocks. Although m_i has been thoroughly studied for other crystalline and sedimentary rocks (e.g., Cai 2010; Richards and Read 2011; Carter 2020), most of these studies focus on deriving m_i from lab data or novel methods for estimating it using rock texture (e.g., porosity). Richards and Read (2011) provide a clear discussion regarding the wide variability of this parameter, and this research highlights that, although all of the tested samples are andesite, they vary considerably in terms of their texture and mineralogy, leading to a wide range of m_i. Villeneuve et al. (2021) and Villeneuve and Heap (2021) show that, for these rock types, where triaxial data are not available, porosity is a potential indicator of m_i (as also shown in Fig. 5j-l), but the wide variation of m_i for a given porosity highlights that m_i depends on several factors, in addition to porosity. As with strength, m_i tends to be higher for fresh rocks than for weathered and altered rocks, which is most pronounced for low-porosity lava (Fig. 11). This shows that more research is required to decipher the links between failure criterion parameters and physical characteristics to improve on empirical data and provide reliable means to estimate these parameters in the absence of triaxial testing, as is often the case in volcano studies.

Fig. 11

Averaged physical and mechanical properties of lava rocks only vs. alteration type and alteration pervasiveness for (a-b) porosity, (c-d) permeability, (e–f) P-wave velocity (Vp), (g-h) uniaxial compressive strength (UCS), (i-j) static Young’s modulus, and (k-l) Generalized Hoek–Brown failure criterion intact rock constant m_i. Note that altered specimens (surface weathering, supergene, and steam heated) have various levels of alteration pervasiveness (minor, moderate, pervasive) and therefore different specimens will shift across categories

The influence of weathering and alteration on physical and mechanical properties

While the extent of altered or weathered minerals is in large part due to fossil hydrothermal systems, we find that lavas are less pervasively weathered/altered than autoclastic breccias and the pyroclastic rock (Fig. 4c). In addition to having more void space, autoclastic breccias and the pyroclastic rock also have more incipient micro-fractures (Mordensky et al. 2018), both of which increase permeability (Fig. 5a) and surface area (Farquharson et al. 2019), creating larger reactive surface areas and enabling the circulation of fluids. Scoria (or pumice in dacitic to rhyolitic environments) fragments large enough to core and study mechanically are rare (del Potro and Hürlimann 2008). Additionally, many pyroclastic rocks are too soft or incoherent to be cored. Thus, this study was limited in pyroclastic sampling, which limits our ability to derive patterns in their physical and mechanical properties.

Fresh rock tends to be less porous and stronger than rock with surface weathering, which is less porous and weaker than rock with supergene or steam-heated advanced argillic weathering (Fig. 4b, Fig. 9a-b, Fig. 11a-b, g-h). A larger percentage of weathering and alteration to clay minerals generally reduces strength and increases the propensity for ductile behavior (Fig. 4c, Fig. 9a-b, e–f, Fig. 11g-h; as also observed by Siratovich et al. 2016; Mordensky et al. 2019a), with the degree/intensity of clay alteration negatively correlating to UCS and Young’s modulus (Fig. 11g-j; as also observed by Watters et al. 2000). Alteration mineralogy is not solely a function of the primary lithology, being also dependent on the local hydrothermal history of the volcano. Thus, there are limitations to assuming that all steam-heated rocks will be weaker than all fresh rocks, as primary porosity and rock history needs to be taken into consideration. In our dataset, this is clear in the strong scatter within the minor and moderate alteration pervasiveness categories, and in surface weathering and supergene weathering categories (Fig. 11g-h).

Some characteristics, such as UCS, static Young’s modulus, and m_i, systematically vary with increasing alteration style and pervasiveness, allowing for a clear distinction between fresh and altered rock (Fig. 11g-l), as also shown in (Heap et al. 2021). Porosity, permeability, and Vp have larger data spreads in weathered and altered rock than in fresh rock (Fig. 11a-f), which also reduces systematic trends through alteration type and pervasiveness. Porosity tends to be higher in altered samples, but more samples in each alteration type and pervasiveness category are required to draw further conclusions (Fig. 11a-b).

Permeability remains within one order of magnitude for all alteration types and pervasiveness, except for supergene and moderate alteration (Fig. 11c-d). Supergene and moderate alteration appear to have a wide range of effects on permeability, although the reason for this is not yet clear due to a small sample number. Additional steam-heated or pervasively altered samples would further clarify the trends seen in this dataset. Pervasively altered samples tend to have the lowest Vp, although a systematic trend through alteration pervasiveness, as observed by other authors such as Pola et al. (2012), is less obvious (Fig. 11e-f).

The relations between physical and mechanical rock properties vary depending on the rock characteristics; for example, porosity may be a better predictor of UCS than alteration type or pervasiveness for rocks with porosity > 10%, whereas alteration and pervasiveness may be better predictors for lavas (which tend to have lower porosity). These trends are affected by high-intensity hydrothermal alteration (Heap et al. 2019b), which develops when rocks are in close proximity to an active hydrothermal system or are exposed to hydrothermal fluid circulation for long periods. This can lead to both primary denser minerals (pyroxene, titanomagnetite) being replaced by less dense secondary minerals (goethite, jarosite), and the precipitation of clay minerals (e.g., kaolinite, montmorillonite) into pores and cracks, resulting in anomalously low porosity, low strength samples (Heap and Violay 2021; Schaefer et al. 2021). This is highlighted by samples RH50 and RH52 (Fig. 6). Sample RH52 is a block of lava taken from the edge of the Crater Lake. The Crater Lake is highly acidic (pH = 0–3) and fluctuates in temperatures (10–60 °C) (Christenson and Wood 1993; Christenson et al. 2010). This sample was subjected to intense alteration and precipitation of clay infilling minerals, particularly in the outer 6 cm, resulting in lower porosity-to-UCS ratios than is typical for lava with low primary porosity (Fig. 4). Sample RH50, a block from the 1995–1996 eruption, represents lava from the upper conduit area within the current vent-hosted hydrothermal system (Kilgour et al. 2010). RH50 has a high percentage of phyllosilicate clay minerals, resulting in anomalously low porosity, low strength, and low magnetic susceptibility (Fig. 4, Fig. 6c, f, i). These anomalous material property relationships are indicative of potentially complex relationships between porosity, mineralogy, and physical and mechanical characteristics of material at depth, which remain to be thoroughly explored.

An additional consideration for material property variations is alteration rate, and spatial and temporal position to aggressive fluids. Alteration by acidic fluids is fairly rapid (Farquharson et al. 2019), and likely faster than surface weathering processes, particularly at high temperatures. For example, samples RH52 and RH42B were both formed in the youngest Whakapapa formation (≤ 15 ky), but RH52 has low porosity yet moderate alteration due to exposure to acidic Crater Lake fluids, while RH42B has high porosity yet remains a fresh sample because it was not exposed to acidic fluids.

Implications for monitoring, modeling, and volcanic hazards

Sample RH50, a lava rock that originated from the upper conduit and was brought to the surface during the 1995–1996 eruptions (see details in Study Area and Materials section), highlights the potential variability of surface from subsurface samples at Ruapehu. The low strength of RH50 due to the high percentage of clay minerals suggests that material in the vent area of Ruapehu (up to 410 m below the Crater Lake; Christenson 2010) could be two times weaker than typical porosity-strength trends of surface samples; RH50 has an average UCS of 93 MPa (Table 3) while the exponential trend suggests a rock with a porosity c. 3.6% would have a UCS c. 190 MPa (Fig. 4). Although we are limited by only having one sample, this does suggest that mechanical properties of the vent area should be carefully considered in numerical modeling. Low apparent magnetic susceptibility (< 0.005 SI), interpreted as a region of altered rock in the current hydrothermal system, has been mapped beneath the Crater Lake extending to ~ 1700 m a.s.l. using an aeromagnetic data inversion (Miller et al. 2020). Other low susceptibility areas have been mapped within the upper cone flanks using a combination of aeromagnetic and hyperspectral surveys and interpreted to be part of older hydrothermal systems (Kereszturi et al. 2020). Thus, larger areas of weak, soft, and altered material may be present throughout Ruapehu’s upper flanks than previously expected, or that is obvious from surface observations. This emphasizes the benefit of using ballistic projectiles (fragments of rock expelled during explosive eruptions, such as block RH50) that have originated from the subsurface to measure the properties in currently active hydrothermal systems (Kennedy et al. 2020). However, subsurface volcanic samples useful for mechanical testing are rare, and those that are acquired may require assumptions regarding sample origin and environment. Thus, determining subsurface material properties, especially at meaningful resolution and under realistic laboratory conditions (e.g., varying confining pressures, temperatures, and saturation), remains a challenge.

This dataset additionally emphasizes that, although trends of weathering/alteration and material properties do exist, properties of volcanic rock can vary significantly within each category and within hand-specimen scales. However, fresh material is typically distinguishable from moderately or pervasively altered rock, which suggests that end-member differentiation is possible for physical and mechanical property mapping. The generally strong control of porosity on other physical and mechanical properties suggests that more accurate relations can be made if a dataset is split by lithology (Fig. 4), which typically have distinct porosities due to emplacement mechanisms. The use of different lithology groups will likely result in physical and mechanical input data at higher spatial resolution than is typically used in geophysical models (Heap et al. 2020).

Implications for spectroscopy-material property correlations

Our data show that, as alteration becomes more prominent, there is a spectral shift from 540 nm to < 450 nm, indicating the occurrence of goethite (430 and 480 nm; Fig. 3a) as a weathering product of hydrothermal pyrite (Kereszturi et al. 2020), and precipitation of native sulfur (390 nm; Fig. 3a). This alteration-induced shift in wavelength location (Fig. 3a) also coincides with alteration-induced changes to material properties (Fig. 11). Using Spearman’s ρ statistical value, Schaefer et al. (2021) demonstrate that these tandem shifts result in strong correlations between laboratory reflectance values and rock properties such as magnetic susceptibility, UCS, porosity, density, friction angle, and static Young’s modulus. The laboratory spectral signatures of carbonate rocks have also recently been shown to correlate with rock strength (Bakun-Mazor et al. 2021). Given the resource- and time-consuming efforts required to characterize rocks in a laboratory setting, future efforts could explore direct correlations of spectroscopy with physical and mechanical rock properties. Shortwave infrared spectroscopy from remote platforms at volcanoes (e.g., Aslett et al. 2018; Kereszturi et al. 2018; Gabrieli et al. 2019) can further support the spatial interpolation of sparsely sampled material data, resulting in large-scale geotechnical maps.

Conclusion

Physical and mechanical properties were determined for a suite of 23 blocks of andesitic lavas, autoclastic breccias, and pyroclastic rocks from Ruapehu volcano in New Zealand with varying weathering and alteration intensities and mineral assemblages. The large number of samples of lavas and autoclastic breccias allowed the distinction of material relations and systemic changes due to weathering and alteration heterogeneities for these rock types. Expanding this research to include volcanic material from other locations (surface and subsurface) will continue to develop these relationships determined herein. Our key findings are:

(1) Porosity has a dominant effect on many physical and mechanical properties. Lavas typically having lower porosities (< 20%) than autoclastic breccias (> 20 %) and tephra (> 40%).

(2) The type of alteration (i.e., weathering or hydrothermal alteration), including alteration intensity, additionally influences physical and mechanical rock properties. The uniaxial compressive strength, static Young’s modulus, and Hoek-Brown constant m_i were found to have more systematic variations with weathering/alteration mineralogy and pervasiveness than variables such as permeability and seismic velocities.

(3) Rocks without weathering or alteration present tend to be less porous and stronger than rock with surface weathering. Rock with surface weathering is less porous, stronger, and less brittle compared to rock with supergene or steam-heated intermediate and advanced argillic weathering.

(4) Vp, density, and Young’s modulus show strong relations with material properties such as porosity and strength.

(5) End-member property differentiation is easily distinguishable (e.g., fresh vs. pervasively altered rock). However, rocks with minor to intermediate weathering or alteration tend to have larger variations in properties due to varying consequences of the nature of the host rock, the fluid type and composition, temperature, and the duration of rock-fluid interaction.

(6) Samples near the vent complex do not follow typical porosity-strength trends as surface samples due to the intensity of clay mineral precipitation and replacement. This has implications for varying the brittle-ductile transition zone, preventing fluid flow, and increasing the likelihood of collapse.

Trends in the pervasiveness of weathering with physical and mechanical properties, along with shifts in the position of spectral absorption peaks as hydrothermal/weathering alteration increases, imply it may be possible to correlate intact rock properties to laboratory or imaging spectroscopy measurements. Extrapolating material properties from imaging spectroscopy is a promising avenue of future research, which could be used to facilitate regional-scale mapping of physical and mechanical properties.