An assessment of hydrothermal alteration in the Santiaguito lava dome complex, Guatemala: implications for dome collapse hazards

A combination of field mapping, geochemistry, and remote sensing methods has been employed to determine the extent of hydrothermal alteration and assess the potential for failure at the Santiaguito lava dome complex, Guatemala. The 90-year-old complex of four lava domes has only experienced relatively small and infrequent dome collapses in the past, which were associated with lava extrusion. However, existing evidence of an active hydrothermal system coupled with intense seasonal precipitation also presents ideal conditions for instability related to weakened clay-rich edifice rocks. Mapping of the Santiaguito dome complex identified structural features related to dome growth dynamics, potential areas of weakness related to erosion, and locations of fumarole fields. X-ray diffraction and backscattered electron images taken with scanning electron microscopy of dacite and ash samples collected from around fumaroles revealed only minor clay films, and little evidence of alteration. Mineral mapping using ASTER and Hyperion satellite images, however, suggest low-temperature (<150 °C) silicic alteration on erosional surfaces of the domes, but not the type of pervasive acid-sulfate alteration implicated in collapses of other altered edifices. To evaluate the possibility of internal alteration, we re-examined existing aqueous geochemical data from dome-fed hot springs. The data indicate significant water–rock interaction, but the Na–Mg–K geoindicator suggests only a short water residence time, and δ18O/δD ratios show only minor shifts from the meteoric water line with little precipitation of secondary (alteration) minerals. Based on available data, hydrothermal alteration on the dome complex appears to be restricted to surficial deposits of hydrous silica, but the study has highlighted, importantly, that the 1902 eruption crater headwall of Santa María does show more advanced argillic alteration. We also cannot rule out the possibility of advanced alteration within the dome complex interior that is not accessible to the methods used here. It may therefore be prudent to employ geophysical methods to make further assessments in the future.


Hydrothermal systems in lava domes
A long-term result of the interaction of water (both meteoric and groundwater) with magma beneath silicic lava domes is the development and persistence of active hydrothermal systems. Within such systems, dome rocks can be altered to clay minerals, significantly weakening portions of an edifice, reducing slope stability and ultimately resulting in failure (Lopez and Williams 1993;Boudon et al. 1998;Reid et al. 2001;Crowley et al. 2003;Zimbelman et al. 2005;Opfergelt et al. 2006;John et al. 2008). At the Santiaguito lava dome complex of Santa María volcano in Guatemala, there is both an evident hydrothermal system and a regional climate characterized by intense bouts of precipitation during the wet season. This study involves an assessment of the degree of hydrothermal alteration using combined analytical and remote sensing approaches. In particular, the objectives are to characterize, to the extent possible, the alteration mineralogy in order to determine the maturity of the hydrothermal system, and to indicate areas of possible concern with respect to potential collapse of old and/or altered dome material.
The presence and assemblage of clay minerals in alteration products is particularly important. Clay-rich materials have the potential to absorb and channel groundwater, as well as increase the fluid pore pressure within an edifice. This in turn can reduce the factor of safety of the slope by promoting the formation of low-strength layers and hence increasing the risk of slope failure (Boudon et al. 1998). Increased pore water hosted in clays in edifice rocks can also create highly mobile cohesive debris flows, which can extend the effects of a sector collapse far beyond the potential runout of dry material (Scott et al. 2001;John et al. 2008).
Flank collapses can occur frequently and repeatedly in volcanic settings which host large, active hydrothermal systems, particularly in stratovolcanoes and silicic lava dome complexes (Vallance et al. 1995;Capra et al. 2002;Clavero et al. 2004). Two pertinent and recent examples are the 1998 debris flow at Casita in Nicaragua (Sheridan et al. 1999;Voight et al. 2002;Opfergelt et al. 2006) and the 1997 debris avalanche at Soufriere Hills, Montserrat Voight et al. 2002). At Casita, an ∼8-ka dacite lava dome complex, a 1.6 million m 3 collapse generating a debris flow on 30 October 1998 was triggered by intense rainfall associated with Hurricane Mitch, and tragically resulted in more than 2,500 fatalities (Sheridan et al. 1999;Kerle 2002). It was found that intense hydrothermal activity had altered the rocks to easily erodible smectite clays (Opfergelt et al. 2006). For the 26 December 1997 debris avalanche at Soufriere Hills Volcano, Montserrat, hydrothermal alteration of the retaining crater wall was implicated as a major contributor to the destabilization and subsequent collapse and depressurization of the lava dome Voight et al. 2002). The resultant debris avalanche deposits contain abundant kaolinite and smectite group clays, alteration products typical of unsealed acid-sulfate hydrothermal systems. In both cases, these low-strength, low-permeability, fines-rich alteration materials were thought to have helped to concentrate water along and lubricate structural discontinuities, thereby reducing the shear strength of the rocks and ultimately leading to destabilization of the edifice Opfergelt et al. 2006).

Santiaguito lava dome complex at Santa María
Santa María, an andesitic stratovolcano, was the site of a devastating and powerful Plinian eruption in October 1902. The eruption lasted for 36 h, expelled more than 7.5 km 3 of material, and left a ∼0.5-km 3 crater in the southern flank of the volcano (Rose 1972a;Williams and Self 1983). The resultant sector collapse scarp remains steep and highly unstable topography; it experiences frequent rockfalls and supplies debris to seasonal lahars and is itself a source of concern with respect to hazards. The focus of this study, however, is on the hydrothermal system associated with the historically and currently active Santiaguito lava domes which began extruding in the base of the crater in June 1922, 20 years after the eruption (Fig. 1a, b). In the 86 years since the dome-building eruptions began, a series of four lava domes have formed, totaling ∼1 km 3 in volume: El Brujo, El Monje, La Mitad, and El Caliente (Fig. 1b) (Harris Fig. 1 a Location of Santa María/Santiaguito in Guatemala. b Aerial orthophoto of Santa María and the Santiaguito dome complex, with dome and river names. The Santiaguito Volcano Observatory (OVSAN) and headquarters of the Finca El Faro are located south of the domes. The extent of the 1929 dome collapse pyroclastic density current, the most extensive dome collapse generated at Santiaguito to date, continued approximately 4 km south of the limit of the image. The location of waters sampled by Walker et al. (2006) al. 2003). Dome growth has migrated from east to west, and has included associated Vulcanian explosions, effusion of a number of ∼1-4 km-long blocky lava flows, and lahars (Rose 1972b;Rose 1973a, b;Rose 1987a;Anderson et al. 1995;Harris et al. 2003). Unlike activity at some other lava dome eruptions (e.g., Sato et al. 1992;Cole et al. 1998;Calder et al. 2002), only relatively small and infrequent dome collapses have occurred at Santiaguito (Sapper and Termer 1930;Rose 1973a). The most extensive one (Fig. 1b), for which there are records, occurred on 2 November 1929, was approximately ≤3 million m 3 , and was likely associated with the locus of active extrusion. Current activity at Santiaguito is focused at the Caliente vent (Fig. 1b), where slow lava extrusion and regular Vulcanian explosions occur at the dome summit (Venzke et al. 2002(Venzke et al. -2012Johnson et al. 2004Johnson et al. , 2008Sahetapy-Engel et al. 2008).
For this study, an understanding of dome structure is important. The summit structure and activity of the currently active Caliente dome is controlled by the slow upward movement of a stiffened plug of dacite in a central conduit leading to both extrusion of a lava flow as well as periodic ash-and-gas eruptions (Bluth and Rose 2004;Johnson et al. 2004;Sahetapy-Engel et al. 2004;Johnson et al. 2008;Sahetapy-Engel et al. Sahetapy-Engel and Harris 2008); the structure of the other lava domes has not been well studied. Sapper and Termer (1930a) made early observations of endogenous growth of the dome complex in the 1920s, but multiple dome units and lava flows have erupted since (Rose 1972b;Rose 1987b;Harris et al. 2003Harris et al. , 2004. A comprehensive geological map of the broader dome complex has been presented by Escobar Wolf et al. (2010), which has provided a foundation for some of our work in determining more detailed dome structure ("Geomorphological mapping of the Santiaguito dome complex" section).
Intense bouts of precipitation have already been implicated as a triggering mechanism for collapses at lava dome eruptions even in the absence of hydrothermal alteration (Matthews et al. 2002;Elsworth et al. 2004;Simmons et al. 2004;Barclay et al. 2006;Matthews et al. 2009). Southwestern Guatemala hosts an extremely wet climate, with rainfall at the summit of Santa María (3,772 m) totaling more than 1,700 mm per year (Walker et al. 2006); precipitation rates at the dome complex have been estimated to be as much as 3,000-4,000 mm per year (Lopez 2004). The 1902 eruption crater of Santa María also drains precipitation directly onto alluvial plains at the foot of the Santiaguito domes ( Fig. 2), which partially runs off in outlet channels to the east and west of the domes, but some of which evidently infiltrates into the base of the domes. The five principle drainage channels below Santiaguito (Fig. 1a-the Conception, Tambor, Nima I and II River valleys, and San Isidro Stream valley) likely receive a significant amount of recharge from the dome complex and crater area (Walker et al. 2006). Given their orientations, if a collapse occurred, the same drainages would direct collapse material or resulting debris flows into populated areas.
The Santa María-Santiaguito system has shown evidence of hydrothermal activity throughout its development. The crater of Santa María produced geysers and steam plumes, and hosted a crater lake for some months following the 1902 eruption (Winterton 1903;Sapper 1904). Once the domes extruded, Rose (1969, 1974) observed both high-temperature (>150°C) and low-temperature (<150°C) fumarole fields. They noted that low-temperature fumaroles were common on blocky talus slopes and that temperatures in the cooler fumaroles were sensitive to seasonal and even daily precipitation, decreasing by as much as 50°C following bouts of rainfall. They also inferred that most of the gas emanating from the coolest fumaroles was from water of a meteoric origin as the ion contents of the fumarolic gases were consistently proportional to the concentrations found in dome lavas.

Geomorphological mapping of the Santiaguito dome complex
An understanding of the geomorphology and internal structural features of lava domes is important for hazard assessment as these features potentially provide planes of weakness which could facilitate collapse. Erosional scarps can expose the interiors of lava flows and lobes which can lead to undermining; ground fractures, vents, and shear lobe boundaries provide surfaces along which slip/slumping can occur. Drainage channels provide pathways to funnel debris flows and/or pyroclastic density currents resulting from those collapses.
Mapping on the Santiaguito dome complex was carried out through field work in February 2010, and based on highresolution aerial orthophotos taken in 2006 by the Instituto Geografico Nacional (IGN) of Guatemala, as well as DEMs generated from aerial photography in 2001 (JICA et al. 2003). Features such as the outline of the main lava flows and major alignments were adapted (in consultation with the authors) from Escobar Wolf et al. (2010). Our focus, however, was on mapping small-scale features such as individual shear lobes, extrusion directions and spines on the dome summits, as well as structures such as crater rims, vents, collapse scars, fumarole fields, and playas (small flatbottomed drainage basins with intermittent standing water; Fig. 3). The locations of erosional scarps and drainage channels were derived directly from the aerial photos.
Discrete extrusive vents, Fig. 3-marked in red, are visible on the summits of the El Brujo and Caliente domes, whereas areas where extrusion clearly occurred along linear zones are present near the summit of El Brujo and on the long axis of El Monje (Fig. 3). A number of arcuate ridges are located on all of the domes, the largest of which are associated with vents on the summits of El Brujo and Caliente; these are interpreted to be remnants of old crater rims. A smaller arcuate feature, interpreted to be a collapse scar, is located on the NW flank of El Brujo above a hummocky deposit which overlies an older lava flow (Fig. 3). Erosional scarps (outlined in green- Fig. 3) highlight areas where materials, such as talus and parts of lava flows, have been removed from the domes. These scarps are often associated, downstream, with established drainage channels (marked in blue- Fig. 3), although the northern side of El Brujo hosts numerous channels that do not have erosional scarps at their sources. Of particular note are the extensive erosional scarps on the SE flank of El Brujo and S flank of El Monje (Fig. 4), which expose layers of talus and the cores of several 10-30-m-thick lava flows.
Fumarole fields are common on all four domes (Fig. 3). There is a large field of low-temperature (<150°C) fumaroles in the saddle between El Brujo and El Monje, as well as isolated fumaroles around the spines on the crest of El Brujo and El Monje (Fig. 5). High-temperature fumaroles, observed with a high-resolution thermal imaging camera, are present only on the summit and high on the south flank of Caliente.

Background
In volcanic settings, hydrothermal alteration occurs by percolating fluids from magmatic degassing, convection and heating of meteoric groundwater, meteoric precipitation, or some combination of all three (John et al. 2008). The circulation of either low-temperature (<150°C) water and/or hotter acidic magmatic gases can promote the alteration of volcanic minerals and glass to clays. In particular, feldspars and volcanic glass, which are abundant in both mafic and felsic volcanic rocks, are especially prone to alteration (Wohletz and Heiken 1992;Velde 1995).
The specific alteration mineral assemblages in volcanic systems (Table 1) are useful indicators of fluid temperature and pH conditions (Utada 1980;Wohletz and Heiken 1992;Velde 1995;Thompson and Thompson 1996;Barnes 1997;Pirajno 2009). Acidic alteration (intermediate and advanced argillic) from the interaction of magmatic gases, particularly SO 2 and H 2 S, with groundwater commonly produces kaolinite and smectite group clays, mainly by replacement of K-feldspar and glass. Higher pH conditions (sericitic/phyllic and propylitic alteration) produce clays and micas such as illite and sericite, as well as biotite, epidote, quartz, chlorite, and minor sericite. Silicic alteration (which produces sinters of hydrous amorphous silica, including opal and chalcedony) indicates the presence of rock dissolution without transformation of minerals into secondary alteration products (Fournier 1985). The location of hydrothermal alteration zones is controlled by structural variations in the host rocks, such as fracture distribution, porosity, and permeability; crater and vent locations; and large-scale features such as collapse scars and contacts between old and new volcanic edifices. All of these influence fluid circulation, and thus location of alteration which commonly occurs along cracks, voids, and grain boundaries, both on small and large spatial scales (Velde 1995). These factors are all particularly relevant to lava dome complexes, which contain complicated networks of fractured and fragmented material such as multiple layers of lava and talus, and rocks of varying porosity and permeability.

Sample analysis
For this study, samples of lava and unconsolidated ash were collected from the interior walls of fumaroles on two  Rose et al. (1970) were also used to map fumarole locations currently inactive domes (El Brujo and El Monje). The samples were collected in February 2010 from nine lowtemperature (<150°C) fumaroles; locations are noted on Fig. 2. Both X-ray diffraction (XRD) and visible to shortwave infrared (SWIR; 400-2,500 nm) laboratory spectral analyses (henceforward referred to as IR spectral analysis) were conducted on the sample suite. In particular, IR spectral reflectance measurements allow for direct comparison with, and ground truth validation of, remote sensing mapping results ("Remote sensing alteration mapping" section). For XRD, the samples were dissolved in deionized water and allowed to settle according to particle size separation methods described in Moore and Reynolds (1997). Airdried and oriented mounts on glass slides were made with a 2-μm clay suspension, also according to methods described by Moore and Reynolds (1997). Analyses were conducted on the Siemens D500 X-ray diffractometer, CuKα, graphite monochrometer at the University at Buffalo. XRD patterns (Fig. 6) were recorded for a scan range of 0-30°2θ with a step size of 0.02°2θ for each sample. Visible to infrared spectral reflectance measurements were made on the same powder and rock fragment samples using a Perkin-Elmer Lambda 900 laboratory spectrometer, with an internal artificial light source. Reflectance values were measured relative to "Spectralon," a high reflectivity white-colored plastic standard.
XRD results for samples 4, 7, and 9 (collected in the saddle between El Brujo and El Monje) are shown in Fig. 6a, c, and, e, respectively, alongside corresponding IR spectral reflectance patterns for the same samples (Fig. 6b, d, and f). All three samples show peaks at 22°2θ and 28°2θ (Fig. 6a, c, and e) which are consistent with a secondary peak and the major peak, respectively, of a feldspar mineral (most likely plagioclase). Sample 7 (Fig. 6c) displays the strongest peak at 28°2θ, suggesting it has the most abundant plagioclase feldspar of the three samples, while sample 9 ( Fig. 6e) appears to contain less plagioclase feldspar relative to the vitric and amorphous phases that dominate the spectra. All three samples display curves with an apex in the 23-24°2θ range which indicate varying amounts of silica (opal-C or cristobalite). To distinguish between the two or determine precise amounts would require a more detailed analyses of the samples than could be undertaken for this study (Hillier and Lumsdon 2008;Hubbard, personal communication). However, the short-order crystalline structure of opal has long since been determined by studying its characteristic XRD patterns (Jones et al. 1964;Jones and Segnit 1971;Graetsch 1994). Amorphous opals such as precious opal, potch opal, and hyalite produce broad XRD peaks centered at approximately 23°2θ (for CuKa radiation). Peak broadening is indicative of decreasing amounts of short-order arrangement (e.g., sample 9- Fig. 6e). Better crystalline varieties (opal-C) give rise to either a sharper peak at 22°2θ (e.g., samples 4 and 7- Fig. 6a, c, respectively), a double peak between 20 and 24°2θ (opal-CT), or quartz peaks at 20.8°2θ and 26.67°2θ (chalcedony). Based Fig. 4 Erosional surfaces of the Santiaguito domes. a View from OVSAN, approximately 5 km south of the domes. Eroded surfaces are delineated by dashed lines, and include deep "barrancas" on the southwestern slopes of the dome complex. b Looking south from the saddle between El Monje and La Mitad at the eroded edge of a lava flow levee on the relative intensities of the combined opal-C+plagioclase peak at 22°2θ (Fig. 6a, c, and e), opal-C abundances appear to decrease from the highest in sample 4 (Fig. 6a), lower in sample 7 (Fig. 6c), and the lowest in sample 9 (Fig. 6e). This is also supported by the IR spectral reflectance patterns shown in Fig. 6b, d, and f which show decreasing yet broad hydrous silica absorption features between 2,120 and 2,360 nm ) as they progress from sample 4 ( Fig. 6b-strongest hydrous silica feature) through sample 7 (Fig. 6d moderate hydrous silica feature) to sample 9 ( Fig. 6f weakest hydrous silica absorption feature).
Additional samples (not shown) display XRD patterns devoid of sharp peaks, and are much shallower and broader in intensity than samples 4, 7, and 9 (Fig. 6). These other samples likely represent amorphous signatures from vitric ash dominating the samples. In the analyses which did show sharp intensity peaks (Fig. 6a, c, and e), none correspond to recognized peaks for clay minerals, which would have been found in the 2.5-15°2θ range (Moore and Reynolds 1997). In these cases, it could be that small amounts of clay and/or sulfate alteration minerals are masked by the strong amorphous signature of volcanic glass. Because of this, secondary electron images were taken using a scanning electron microscope (SEM) in the unconsolidated ash and backscattered images (BSE) were collected from the thin sections, which could not be re-examined later by XRD. Additional qualitative measurements were conducted using energy dispersive X-Ray (EDS) analyses while the images were taken.
Samples of air-dried ash were placed on non-conducting, double-sided carbon-coated tape for secondary electron imaging. Double-polished thin sections were created from rock  Information summarized from Utada (1980), Wohletz and Heiken (1992), Velde (1995), Thompson and Thompson (1996), Barnes (1997), and Pirajno (2009) chips and were carbon-coated prior to analysis. Analyses were conducted on the Hitachi S-4000 Scanning Electron Microscope at the University at Buffalo South Campus Instrumentation Center. BSE images were collected at 20.0 kV, medium beam for thin sections, while secondary electron images were acquired of the ash samples. Qualitative chemical analyses were obtained using the EDS probe on the microscope. Secondary electron images of ash (Fig. 7a, b) revealed blocky, equant grains in the 100-200-μm size range, with abundant adhering grains in the 0.5-10-μm range. EDS measurements showed the larger grains to be mainly dacite glass or dacite groundmass, but some of the smaller grains evidenced chemistries consistent with montmorillonite clay. No well-developed crystalline forms were present, however, suggesting that the clay minerals were either amorphous or in the early stages of development (e.g., allophane and imogolite). Feldspars and void spaces in the thin sections contained abundant particles in the <10-μm size range (Fig. 8a, b), but while EDS sampling of these particles again returned clay and zeolite mineral chemistries, no image showed easily identifiable crystalline forms of the minerals.

Background
Remote sensing using multi-and/or hyperspectral image data is particularly useful for mapping surface hydrothermal alteration in remote and inaccessible locations (Crowley and Zimbelman 1997) and therefore expands on the coverage obtained from the direct sampling. Clays and hydrous sulfate minerals (such as kaolinite and alunite) show distinct spectral features in the SWIR region, which spans 1,000-2,500 nm in wavelength. Spectral features in these minerals are related to vibrations of structural hydroxyl groups and molecular water (Hunt and Ashley 1979). The Advanced Spaceborne Thermal Emission and Reflection Radiometer  Fig. 6 Powder X-ray diffraction analyses (a, c, and e) and IR spectral analyses (b, d, and f) of samples of ash on glass slide mounts. All samples show a strong-to-weak curve indicating the presence of hydrous silica and plagioclase feldspar.
Plagioclase is abbreviated to plag in mineral peak labels (ASTER) is a multispectral imaging instrument aboard NASA's Terra satellite, which also carries the Moderate Resolution Imaging Spectroradiometer (MODIS). ASTER operates in 14 bands, with 3 bands in the visible and nearinfrared (VNIR) region (0.52-0.86 μm), 6 bands in the SWIR region (1.6-2.43 μm), and 5 bands in the thermal infrared region (8.125-11.65 μm), the latter of which are not used in this study. The MODIS sensor collects data on atmospheric moisture concurrently with ASTER scenes (Gao and Kaufman 1998) which are used to correct for atmospheric effects and for generating high-level ASTER image processing products (Mars and Rowan 2010;NASA 2004NASA , 2011. Two images from the Hyperion spaceborne imaging spectrometer, on NASA's EO-1 satellite, are also used. Hyperion provides 242 spectral bands spanning 0.4-2.5 μm in wavelength, with 172 channels in the critical SWIR region where diagnostic spectral absorption features related to clay, sulfate, and hydrous silica alteration minerals are most prominent   (NASA 2008). The VNIR bands of the ASTER scenes (15 m resolution) were resampled to SWIR (30 m) resolution in the ENVI image analysis program (ITT 2010), and the ACORN processing utility (ImSpec 2004) was used with Terra MODIS atmospheric moisture data to conduct atmospheric correction and calibration of "at sensor" radiance values to scaled reflectance from the surface. The Hyperion scenes (also at  30 m spatial resolution) were subset from 242 to 196 bands to eliminate repeating and unused bands. They were also calibrated to reflectance using ACORN, which corrects for atmospheric moisture on a pixel by pixel basis (e.g., Gao et al. 1993), as well as the column by column radiometric shifts in wavelength caused by the instrument's pushbroom engineering design (Folkman et al. 2001).
Our alteration mineral mapping procedures follow more detailed methods outlined in Hubbard et al. (2003), but are briefly described here as follows. Representative image spectra spanning the visible and near infrared (VNIR, 400-1,000 nm) and SWIR bands of both our ASTER and Hyperion scenes were determined. The resulting endmember spectra (Fig. 9a, b) were chosen using the Pixel Purity Index (PPI) method (Boardman et al. 1995) in order to locate the most homogeneous (or least mixed) targets. Prior to PPI analysis, vegetation and thermal hot spots (<150°C) were eliminated by masking, though dry vegetation lacking diagnostic chlorophyll spectral absorption features can still interfere with the identification of certain sericitic alteration minerals (van Ruitenbeek et al. 2006). The ASTER endmembers (Fig. 9a) and the Hyperion endmembers (Fig. 9b) were identified based on visual inspection and by comparison with reference spectra from the USGS mineral spectral library (Clark et al. 1993). A spectral matched-filter algorithm (Harsanyi and Chang 1994) and appropriate threshold values were then used to map the spatial distribution of pixels that best matched the characteristic shapes of the reference spectral endmembers (Fig. 9).
The resulting mineral maps derived from the Hyperion and ASTER images reveal the presence of mainly hydrous silica alteration on exposed erosional surfaces of the domes (Fig. 10a, e, and f). Interestingly, ASTER (Fig. 10a, b, d) and Hyperion (Fig. 10c) maps also show mixtures of alunite, kaolinite, and hydrous silica (advanced argillic alteration) in the upper part of the 1902 eruption crater headwall scarp on the Santa María edifice. The ASTER image from 8 December 2000 (Fig. 10a) has the best atmospheric calibration conditions (i.e., lowest humidity), and also shows hydrous silica on the erosional scarp between the La Mitad and El Monje domes, as does the most current image (Hyperion from 22 January 2008, Fig. 10f). These hydrous silica minerals may be due to opal-A (hyalite), opal-C as suggested by our field and laboratory results, or perhaps even chalcedony, though the latter mineral and areas possibly containing it could not be accessed for sampling. Lesser amounts of argillic and phyllic mineral assemblages also appear to be present in scarps on the domes. However, hydrous silica minerals are more prevalent in the best-calibrated images (e.g., ASTER from 8 December 2000- Fig. 10a and Hyperion from 22 January 2008- Fig. 10f). Pixels that are mapped in some images, but not in others, are likely due to ash-cover from eruptions in between image dates, and/or poor calibration due to atmospheric water vapor, ash, and haze from fumaroles on the Caliente dome in the vicinity of circled hot spot areas (Fig. 10, all images). In addition, the general lack of mineral signatures on Caliente may also be due to the ash cover produced by frequent eruptions; this ash is more likely to persist on Caliente simply because it is often dispersed away from the other domes, and the ash that does fall on those domes is probably at least partially removed by seasonal rainfall.

Aqueous geochemistry
Background Inferences about the origin and evolution of hydrothermal waters can be made by comparing their 18 O/ 16 O and deuterium/hydrogen (D/H, or 2 H/ 1 H) ratios with that of local meteoric (unequilabrated) water. Deviations from the Meteoric Water Line (MWL) of Taylor (1979), represented by the equation  (Taylor 1979). The shift may be as small as 1 or 2‰ (e.g., in Fig. 9 a End member spectra for alteration minerals in ASTER 6-3-1 RGB images; profiles are offset for clarity. b End member spectra for minerals in Hyperion 2205-810-560 nm RGB images. Spectral profiles are offset for clarity; "bad bands" have been masked and appear as gaps in the profile. Note that these do not represent all of the spectra used for mapping, but are rather a subset of the best spectral patterns available for these minerals. Individual scenes were mapped using their own spectral end-members, chosen with the PPI method volcanic areas with a higher water/rock ratio), or as much as 15‰ (e.g., in very hot waters circulating through high-18 O sediments). D/H values are not greatly affected by this exchange because the rocks contain little initial hydrogen compared to the amounts of the water involved. A δ 18 O shift not only indicates that waters remain in the hydrothermal system long enough to exchange isotopes with the host rock but also increases the likelihood of alteration occurring as the waters equilibrate. This may be further substantiated with Na-K-Mg geoindicators (Giggenbach 1988), which may be used to compare the relative cation contents of hydrothermal waters to known hydrothermal waters which have been classified as old (equilibrated with their surroundings) or young (not equilibrated with their surroundings).
A regional study of spring water geochemistry was undertaken by Templeton (1999) and later published by  Walker et al. (2006). Their work detailed a sampling campaign conducted on numerous hot and cold springs in the vicinity of Santa María, as well as within the nearby Zunil geothermal field (located 7 km to the NE of the volcano). That study's goal was to understand regional provenances and characterizations of the hot springs through their aqueous geochemistry and stable isotope concentrations. Springs in the vicinity of Santa María and Santiaguito have a dominantly bicarbonate character and high Na + contents; Walker et al. (2006) interpreted this to be a result of interaction between spring water and rocks of the Santiaguito dome complex, which contain elevated amounts of sodium compared to other regional volcanic units (Rose 1987a;Walker et al. 2006). Recharge elevations calculated from stable ion ratios indicate that the springs are mainly supplied by surface runoff from Santa María, and stable isotope trends indicate that the water in the springs is mainly meteoric in origin (rather than magmatic). Although Santa María and Santiaguito are relatively close to the Zunil geothermal field, δ 13 C ratios suggest that the two hydrothermal systems are independent, and spring waters in the vicinity of Santiaguito and Santa María are affected solely by shallow processes, such as those within the edifice (Walker et al. 2006).

Re-examination of local aqueous geochemistry
Here, we re-examine the geochemical data of Walker et al. (2006) from the perspective of alteration processes at Santiaguito, focusing on the data collected from nine springs in drainages below the dome complex (Fig. 1a). Cation contents in water were plotted in terms of total content and compared with meteoric and surface waters using the methods of Giggenbach for determining cation geoindicators (Giggenbach 1988). Hot spring waters around Santiaguito show an appreciable increase in major cations when compared to meteoric and surface runoff values (Fig. 11). In particular, Na, Ca, and Mg show the most significant increases; meteoric values for these cations are approximately zero, while surface runoff shows only a slight increase in ion content. There is a general trend of cation content decreasing with distance from the domes. Stable isotope ratios (δ 18 O vs. δD) of the waters around Santiaguito were compared with the MWL, δD08×δ 18 O+10) to determine if there is an established geothermal system within the domes. For example, waters rich in Cl and acidic geothermal waters tend to show a significant shift toward higher δ 18 O values. This increase results from isotopic exchange with country rocks having initially high δ 18 O values (Hoefs 2004). A comparison of the stable isotope ratios for 18 O/ 16 O (δ 18 O) and D/H (δD) of these waters shows a slight shift away from the meteoric water line in those waters closest to the domes, but only ∼0.3-1.0‰ increase in δ 18 O (Fig. 12). When cation concentrations for the Mg-Na-K system are plotted on Giggenbach's (1988) geoindicator diagram, they fall firmly in the immature waters field, as well as very near the meteoric water value (Fig. 13).  Fig. 11 Geochemical analyses of major cations in meteoric water, springs, and a surface stream near the Santiaguito dome complex. Locations (not including meteoric water) are ordered by increasing distance from the dome complex. In locations where multiple water samples were taken from the same spring, data are averaged; at Santiaguito springs 1, 2, and 3, samples were taken from separate springs and are thus treated as separate data sets. Data for figure from Walker et al. (2006) Walker et al. 2006). MWL is the local meteoric water line (δD0 8×δ 18 O+10). The analyses show no shift toward increased 18 O, which is characteristic of hydrothermal systems containing alteration minerals (Taylor 1979)

Discussion
Our remote sensing analyses are summarized together with fumarole and spring locations on an oblique DEM with a superimposed slope map (Fig. 14). Areas with slopes >40°a re indicated, and zones rich in alteration often coincide with such slopes. This figure provides an overall framework for considering stability across the whole Santiaguito-Santa  Fig. 13 Ternary diagram of Na/ 400, K/10, and √Mg after Giggenbach (1988). Concentrations representing full water-rock equilibrium at temperatures from 20 to 300°C are indicated by the "Full Equilibrium" line. Santiaguito waters plot as "immature." Figure modified from Walker et al. (2006) Fig. 14 3D oblique view of the Santiaguito complex and Santa María volcano. Colors represent areas of high slopes (>40°, in red), outlines of areas where alteration minerals are present (aqua), fumarole fields (yellow) and springs (stars). Zones of alteration were drawn from the combined geographical distribution of all alteration mineralogies determined from our remote sensing data ( Fig. 10) but access for the purpose of direct sampling is too hazardous. Remote sensing results imply any surficial acidic alteration here is also minimal. Fumarolic activity on Caliente's summit is largely concentrated around the margins of active extrusion (Bluth et al. 2002, field observations) and on the northern part of the outer crater rim, but resurfacing by frequent deposition of fresh ash may also obscure bedrock signatures. The western domes have not erupted since the 1970s, and as such, the magmatic signature of their fumaroles might be expected to be significantly lower than those on Caliente. Meteoric water is likely an important component of these fumaroles; in this location, a shallow hydrothermal system dominated by meteoric water flowing through hot rock would lack the acid-sulfate components (from magmatic gases and fluids) that would enhance the effects of low-temperature alteration.
The dissolved ionic contents of springs draining the base of the Santiaguito dome hydrothermal system indicate significant water-rock interaction in the form of rock dissolution, but little isotopic exchange with the host rocks. Taylor (1979) indicated that while the δ 18 O shift caused by isotopic exchange (i.e., alteration mineral formation) was likely to be small, it would be on the order of at least 1-2‰. The δ 18 O shifts observed at Santiaguito are at most 0.5‰. We suggest that these slight shifts reflect residence times that are too short for the water to equilibrate with its host rocks, although dissolved ion contents indicate that some dissolution does occur. The Na-K-Mg geoindicator ternary diagram of Giggenbach (1988) corroborates this: The waters at Santiaguito fall in the immature field, indicating that they have not had sufficient time to fully or even partially equilibrate with the geothermal reservoir.
Meteoric water percolating through the ground to the water table would undergo heating and evaporation, which concentrates ions at ratios equal to that of meteoric water. This water flows a short distance through the domes within a few days to weeks (which would account for the lack of further water-rock interaction). Barnes (1997) states that equilibrium between fluids and silicate minerals is generally achieved on a time scale of weeks to months rather than days, which further supports the idea of a short water residence time involved here. The quantity of water moving through the domes, or the spring flux, was not measured, but may also be important. For example, Taylor (1979) suggests that the small isotopic shifts may be characteristic of volcanic areas with a high water-rock ratio. However, without knowing the spring flux and δ 18 O values of the dome lavas, it is impossible to verify whether a high water-rock ratio is inhibiting internal alteration.
Another factor that should be considered when discussing the apparent lack of alteration signatures in the springwater is the limited area that is drained by the hot springs sampled. The hot springs are clustered below the eastern part of the dome complex, mainly below Caliente. If their only inputs come from water draining through Caliente, then we can only say with certainty that Caliente is not likely to be experiencing internal hydrothermal alteration. The state of the interior of the other domes has yet to be determined; and because the inactive domes lack persistent drainage from springs, other geophysical methods would have to be employed to assess this.
The spectral mapping using ASTER multispectral and Hyperion hyperspectral satellite images does indicate the presence of hydrothermal alteration minerals on erosional surfaces of both the Santiaguito dome complex and the Santa María collapse scar headwall. Two different alteration mineral assemblages were mapped: (a) kaolinite and alunite mixtures in the summit area of the Santa María collapse scar related to advanced argillic alteration from acid-sulfate-rich fluids, and (b) steam-heated, fumarolic alteration dominated by hydrous-silica minerals such as opaline-silca (and perhaps chalcedony), which is consistent with low-temperature silicic alteration in the domes. These data support the geochemistry-based conclusion that the main hydrothermal fluid input to the domes is meteoric, and that minerals which are found in the XRD and SEM imaging of dome lava and ash samples are largely the result of dissolution of dome rock and precipitation of hydrous silica, with minimal to no clay mineral (e.g., montmorillonite) formation.
The prevalence of alunite and kaolinite at the summit of Santa María may indicate a once active magmatic-hydrothermal system dominated by acidic gases. We are not currently aware, however, of the existence of active fumaroles associated with the summit or eruption scar of Santa María. A self-potential and soil CO 2 survey conducted by Bennati et al. (2011) of the area around Santa María (but not extending to the Santiaguito dome complex) was interpreted to indicate the potential presence of a hydrothermal system, related to a residual magmatic body from the 1902 eruption, extending radially about 800 m from the summit of Santa María. Additional kaolinite-mapped pixels within the Santa María scar and downslope of the domes may be due to small debris flows, alluvium, and colluvium originating from the kaolinite-rich source areas exposed along the headwall summit scarp of Santa María. Wohletz and Heiken (1992) and Lynne et al. (2008) note that opal sinters may be formed when the temperatures in hydrothermal waters drop below 100°C, and hydrous silicate minerals can then precipitate out of solution. We observed hydrous silicate minerals on erosional scarps of the dome complex, which would be consistent with water dissolving silica from dome rock and then cooling as it percolates out to the surface. The precipitation of siliceous sinter, in sufficient quantities, may have implications for dome stability and/or dome permeability (Boudon et al. 1998;Komorowski et al. 1997;Komorowski et al. 2010). Based on our work, however, we cannot determine whether sinter at Santiaguito is sufficiently pervasive to have such effects.
Although the original goals of this study were to map alteration on the Santiaguito dome complex, the remote sensing evidence of acidic alteration on the summit of Santa María and its implications should be discussed. The 1902 eruption scar is already known to be a site of active and continuous erosion, and even a cursory look at the morphology of the summit of Santa María shows very steep, and in some places vertical cliffs (Fig. 14). The remote sensing evidence for acidic alteration and clay minerals and the precipitous topography in this area leads us to conclude that rockfalls from the scarp, some of significant volume, are likely a more immediate risk than collapses from the domes themselves. The analytical data collected here did not extend to the Santa María scarp, so the remote sensing data that indicate the presence of advanced argillic alteration require ground truthing. Confirmation of the presence of extensive regions of alteration combined with oversteepened topography as well as environmental agitators such as intense precipitation and/or regional earthquakes would clearly indicate that collapses from the upper scarp region present significant risk. Monitoring of the conditions related to deformation and morphological evolution of this region should be implemented, including simple and regular checks for ground cracks. Future work might also include determining volumes of altered material on Santa María, which could then be used in forward models to predict the impact of a collapse of the summit scarp on downslope regions (Hubbard 2001;Crowley et al. 2003).

Conclusions
While the geochemical analyses conducted in this study provide evidence that the Santiaguito dome complex does not host pervasive hydrothermal alteration, there is some evidence in our remote sensing results for limited and perhaps localized presence of clay minerals on the dome surfaces. The lack of evidence for more pervasive alteration (i.e., advanced argillic alteration similar to that on Santa María) on the domes is somewhat curious, given the feldspathic lavas and wet, low-temperature conditions, which could favor clay mineral formation. This certainly does not, however, preclude the possibility that internal alteration is occurring that could contribute to the instability of the lava domes, nor does it rule out collapses triggered by other processes such as exogenous dome growth or endogenous injection of new magma, or regional seismic activity.
The spectral end-member mapping results were somewhat limited by cloud-cover, surface ash-cover, atmospheric humidity issues, gases from active fumaroles, and haze from both wildfires and the active Caliente dome. These conditions are common problems when using satellite imagery to map minerals at volcanoes in tropical and (or) developed areas. However, persistent observations of the same area (in this case, over an 8-year period) can still yield useful information about hydrothermally altered bedrock by catching rare, but ideal conditions needed for clear imagery and good exposures of altered rocks beneath ash, vegetation, or snow cover. In the case of Santiaguito, multiple images reveal reasonably consistent information about alteration mineral types and locations, which can be tested through on-theground sampling. Geochemical data from ash and lavas around the fumaroles confirm the presence of silicic sinter on the domes, while clay and sulfate minerals indicative of a higher degree of potential edifice-collapse hazard are mainly observed within the summit scarp of Santa María.
The aqueous geochemistry of springs around the base of the domes seems to indicate that while water-rock interaction (e.g., dissolution and precipitation of silica) is occurring, clay mineral formation is unlikely in the area of the dome complex that is drained by the hot springs sampled in Walker et al. (2006) and Templeton (1999). Clustering of the spring water geochemical data in the immature field suggests that the meteoric waters supplying these springs spend relatively little time within the domes. Also, with their nearneutral pH values, this deprives the dome hydrothermal systems of two important factors: acidity and long water residence time, either of which would contribute to more intense hydrothermal alteration.
The limited area drained by the springs sampled precludes us from saying conclusively that there is no hydrothermal alteration occurring within the domes. It may be possible that internal alteration is occurring, but not in areas that drain into the hot springs. In order to assess this, future geophysical surveys such as electrical resistivity/conductivity, self-potential, or magnetotellurics (Fuji-ta et al. 1999;Finn et al. 2001;Nicollin et al. 2006;Bedrosian et al. 2007;Finn et al. 2007;Komori et al. 2010;Lesparre et al. 2012) would be necessary. Finally, and of some concern, is that the results found here indicate that the precipitous cliffs near the headwall of the Santa María scarp are additionally compromised by the indication that argillic alteration is prevalent there. The region is prone to seismic hazards and precipitation rates which could both diminish stability and trigger collapses from the scarp headwall. We suggest that it would be useful to focus further work on assessing hazards related to such a collapse.
(University at Buffalo) provided analytical support for the XRD analyses, and Peter Bush donated time on the UB South Campus Instrumentation Center SEM. Finally, the authors wish to thank P. Delmelle, JC Komorowski, J. Mars (USGS), and an anonymous reviewer for their comments and suggestions for improving the manuscript.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.