Cofre de Perote volcano is a compound, shield-like volcano located in the northeastern Trans-Mexican volcanic belt. Large debris avalanche and lahar deposits are associated with the evolution of Cofre. The two best preserved of these debris-avalanche and debris-flow deposits are the ∼42 ka “Los Pescados debris flow” deposit and the ∼11–13 ka “Xico avalanche” deposit, both of which display contrasting morphological and textural characteristics, source materials, origins and emplacement environments. Laboratory X-ray diffraction and visible-infrared reflectance spectroscopy were used to identify the most abundant clay, sulfate, ferric-iron, and silica minerals in the deposits, which were either related to hydrothermal alteration or chemical weathering processes. Cloud-free Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) remote sensing imagery, supporting EO-1 Hyperion image spectra, and field ground truth samples were used to map the mineralogy and distribution of hydrothermally altered rocks on the modern summit of Cofre de Perote. The results were then compared to minerals identified in the two debris-avalanche and debris-flow deposits in order to assess possible source materials and origins for the two deposits.

The older Los Pescados debris-flow deposit contains mostly halloysite and hydrous silica minerals, which match the dominant mineralogy of soils and weathered volcanic deposit in the surrounding flanks of Cofre de Perote. Its source materials were most likely derived from initially noncohesive or clay-poor flows, which subsequently bulked with clay-rich valley soils and alluvium in a manner similar to lahars from Nevado del Ruiz in 1985, but on a larger scale. The younger Xico avalanche deposit contains abundant smectite, jarosite, kaolinite, gypsum, and mixed-layered illite/smectite, which are either definitely or most likely of hydrothermal alteration origin. Smectite in particular appears to be the most abundant and spectrally dominant mineral in summit ground truth samples, ASTER mapping results, Xico avalanche deposit, and an older (pre-Xico avalanche) deposit derived from collapse(s) of ancestral Cofre de Perote edifice. However, both Xico avalanche and Los Pescados debris flow deposits show some evidence of secondary, postemplacement weathering and induration, which is evident by the presence of gibbsite, and hydroxyl interlayered minerals, in addition to recently formed halloysite and hydrous silica (i.e., indurating) cements. Field-based, visible infrared image spectroscopy (VIS/IR) spectral measurements offer the possibility of distinguishing primary minerals of hydrothermal alteration origin in debris-avalanche and debris-flow deposits from those produced either by in situ chemical weathering or bulked from weathered source materials.


Volcanic debris avalanches and debris flows (i.e., lahars) originate as slope failures high on volcanic edifices and transport enormous volumes of fluidized rock and soil into surrounding river valleys. In populated areas, these phenomena can be incredibly destructive and deadly, as tragically exemplified by the >22,000 people killed by lahars from the Nevado del Ruiz volcano, Colombia, in 1985, during a minor eruption that melted glacial ice and bulked with clay‑rich soils and alluvium from the surrounding valleys (Lowe et al., 1986; Pierson et al., 1990). Lahars containing abundant hydrothermally produced clays, such as the 5.6 ka Osceola mudflow from Mount Rainier (Crandell, 1971; Scott et al., 1995; Vallance and Scott, 1997) and the 16.4–16.6 ka Teteltzingo lahar from Citlaltépetl (Carrasco-Núñez et al., 1993, 2006), have left extensive deposits many tens of kilometers from their sources, signaling that population centers situated far from a volcanic edifice may still be at risk. Volcanic debris avalanches and debris flows are not always accompanied by eruptive activity. They can be triggered by a variety of factors such as increased precipitation, as exemplified in 1998 by the Casita avalanche and lahar triggered by Hurricane Mitch (Sheridan et al., 1999; Scott et al., 2005); slope instability and over steepening caused by glacial erosion (Crowley et al., 2003); strong seismic activity (Martinez et al., 1995; Scott et al., 2001); or they can occur even without warning.

A number of studies have made interpretations about the origins of ancient volcanic debris avalanches and flows based on the type and distribution of clay, sulfate, and silica minerals indicative of either hydrothermal alteration and/or chemical weathering processes (e.g., Crandell, 1971; Carrasco-Núñez et al., 1993; Vallance and Scott, 1997; Vallance, 1999; Capra and Macias, 2000; Capra and Macias, 2002; Pulgarin et al., 2004; Carrasco-Núñez et al., 2006; Murcia et al., 2008). However, few studies have mapped the distribution of clay-rich, hydrothermally altered rocks on volcanoes with downstream populations at risk (e.g., Crowley and Zimbelman, 1997; Hubbard, 2001; Finn et al., 2001; Crowley et al., 2003; Finn et al., 2007) or compared the mineralogy of actual debris-avalanche and debris-flow deposits with those of potential source rock areas on the volcano (e.g., Pevear et al., 1982; Frank, 1983; Hubbard, 2001; Opfergelt et al., 2006), assuming that such rocks indeed still remain on the edifice. For example, Pevear et al. (1982) notes that the 18 May 1980 debris avalanche deposit from Mount St. Helens lacks acid-sulfate minerals such as kaolinite and alunite, but contains abundant chlorite, mixed layered chlorite/smectite (i.e., corrensite), and saponite (an Mg/Fe2+ or trioctahedral smectite), indicative of a sealed hydrothermal system that prevented acidic fluids from reaching the surface or near surface oxidizing environments. Minor amounts of acid-sulfate alteration minerals such as kaolinite, alunite, cristobalite, tridymite, and opal were known to exist locally around the vicinity of fumarolic and geothermal areas on the pre-1980 Mount St. Helens summit dome and goat rocks dome (Pevear et al., 1982 and references therein). These dome rocks were subsequently removed during the 18 May 1980 rockslide, though (not surprisingly) their alteration products are not evident in the resulting debris-avalanche deposit as shown by Pevear et al. (1982). In contrast, Hubbard (2001) used airborne visible–infrared imaging spectrometer (AVIRIS) hyperspectral data to map a variety of hydrothermal alteration minerals on the modern edifice of Citlaltépetl, as well as the remnants of two ancestral edifices, and compared them with the mineralogy of the debris-avalanche and debris-flow deposits that resulted from their collapse.

This study focuses on Cofre de Perote volcano, one of the main volcanoes of the Citlaltépetl–Cofre de Perote volcanic range, which has produced two large (i.e., >108 m3) debris-avalanche and debris-flow deposits dated at ∼42 ka and ∼11–13 ka, informally named and referred to as Los Pescados debris flow and Xico avalanche, respectively (Carrasco-Núñez et al., 2006). Both deposits contain abundant clay and related nonclay minerals and exhibit contrasting morphological and textural characteristics, which provide clues to their different origins and modes of emplacement.

Also, the modern-day summit area of Cofre de Perote volcano contains visible collapse scarps, evidence of glacial erosion, and hydrothermally altered areas. Because chemical weathering and soil-forming (i.e., pedogenic) processes play a critical role in triggering landslides, debris avalanches, and debris flows at tropical latitude volcanoes in particular, we discuss these processes further in the context of clay-mineral formation and the origins of the Los Pescados debris-flow and Xico avalanche deposits. In particular we examine the nature and origin of clay, silica, sulfate, and ferric-iron minerals in the Los Pescados debris-flow and Xico avalanche deposits by comparing them to minerals found within hydrothermally altered rocks collected from possible source areas near the Cofre de Perote volcano summit and mapped imagery using ASTER. In the process, we present field, laboratory, and remote-sensing methods that can be used to study other debris-avalanche and debris-flow deposits of unknown origin at other volcanoes around the world, and to assess the significance of hydrothermal alteration and/or soil-forming processes in generating future debris avalanches and lahars from Cofre de Perote volcano.


Cofre de Perote is the northernmost and second highest (∼4220 masl) volcano of the Citlaltépetl–Cofre de Perote volcanic range. It is situated at the eastern end of the Trans-Mexican volcanic belt (TMVB; Fig. 1). The Trans-Mexican volcanic belt is a Neogene volcanic arc characterized by its oblique geometry with respect to the Middle American subduction zone trench. It contains a wide range of volcanic structures, including large silicic calderas, andesitic stratovolcanoes, silicic domes, and large basaltic monogenetic fields (Demant, 1978). The easternmost Trans-Mexican volcanic belt comprises the Citlaltépetl–Cofre de Perote volcanic range and the Serdán-Oriental Basin, marked by a bimodal volcanism with numerous maars, domes, and cinder cones comprising a scattered monogenetic field. Polygenetic volcanism is mostly limited to the Citlaltépetl–Cofre de Perote volcanic range, where the only remaining currently active volcano in the range is Citlaltépetl (C, Fig. 1). Another important structure to the evolution of the Cofre de Perote volcano and NW of it is the Los Humeros caldera (8, Fig. 1), which is actually composed of three nested calderas, two of which (Los Humeros and Los Potreros) are nested inside an older ancestral caldera.

Cofre de Perote volcano is one of the largest structures of the Citlaltépetl–Cofre de Perote volcanic range. It represents a major physiographic and orographic divide between the Coastal Plain to the east and the basin of Serdan-Oriental to the west, the latter of which forms an altiplano plateau. The difference in elevation between the two physiographic provinces is more than 1200 m, which has been partly attributed to the tilting of pre-volcanic basement rocks (Carrasco-Núñez et al., 2006 and references therein). Differences in climate, vegetation, and rock weathering regimes between the two physiographic provinces are also striking. The western side is a semiarid desert with low (<400 mm) annual precipitation, an eight-month dry period, and annual temperature variations up to 18 °C with an average between 11 and 14 °C (Dubroeucq et al., 1998). Soils on the western side of Cofre de Perote volcano are andosols modified by aeolian processes and caliche (i.e., calcrete) cementation (Dubroeucq et al., 1998). The eastern side of Cofre de Perote volcano has a subtropical to tropical climate with a brief (less than four months) dry season from January to April, and receives >1400 mm of annual precipitation (Elsass et al., 2000). Soils on the eastern flanks are also andosols but are typically indurated with hydrous silica minerals (i.e., silcrete) (Elsass et al., 2000). Land use on the east side of Cofre de Perote volcano favors a dense vegetation canopy ranging from cloud forest to low jungle and savannas with numerous scattered sugar cane and coffee plantations. On both sides of Cofre de Perote volcano, the weathering zone extends to a maximum depth of ∼4 m (Dubroeucq et al., 1998; Elsass et al., 2000), although this varies with slope and often grades into saprolite on the western side (Elsass et al., 2000).

Morphologically, Cofre de Perote volcano could be described as a compound, shield-like volcano with a broad and gently sloping profile (Fig. 2) (Carrasco-Núñez et al., 2010). Considering that Cofre de Perote volcano is probably extinct because its last eruptive episode ended ∼200 ka (Carrasco-Núñez et al., 2006), detailed geologic mapping and study of its potential hazards have received little attention until recent times. Nevertheless, the young-looking scarps along the eastern portions of the summit (Fig. 3) suggest collapse episodes at a time later than the cessation of its eruptive activity. This geomorphological evidence, along with the two documented Los Pescados debris-flow and Xico avalanche and debris flow deposits on the eastern lower slopes (Carrasco-Núñez et al., 2006, 2010), are further evidence of postconstructional collapse and edifice instability. Cirques and U-shaped valleys dominate the western flanks of Cofre de Perote volcano (Fig. 3) and provide evidence of late Pleistocene glacial erosion, similar to that of cirques found on the western flanks of Citlaltépetl and Las Cumbres farther south along the Citlaltépetl–Cofre de Perote volcanic range (Lorenzo, 1964; White, 1986; Heine, 1988; Siebe et al., 1993; Lachniet and Vázquez-Selem, 2005; Rodríguez, 2005).

Carrasco-Núñez et al. (2010) propose that the geologic evolution of Cofre de Perote volcano can be divided into four different stages, three of which correspond to major effusive constructional stages, and a fourth corresponding to a collapsing stage as depicted in Figure 4. Eruptive activity at Cofre de Perote volcano varies in age with the earliest activity dated at 1.3 ± 0.12 Ma (Pleistocene) for lavas exposed east of the city of Coatepec on the eastern flanks, while on the western flank, lavas are dated at 0.51 Ma (Fig. 4A). This first-stage effusive activity is followed by a second stage comprised of superimposing lava flows and domes, which change from basaltic-andesitic to andesitic-dacitic in composition. These eruptive products have been dated within the range of ages from 0.31 to 0.42 Ma (Fig. 4B). The third and final effusive stage lasted from ∼0.25 Ma until eruptive activity ceased at ∼0.2 Ma (Fig. 4C) (Carrasco-Núñez and Nelson, 1998). Figure 5 shows the distribution of the geologic units representing the main effusive stages, as well as debris-avalanche and debris-flow deposits representing the fourth stage of Cofre de Perote volcano, corresponding to Los Pescados debris-flow DF and Xico avalanche (Fig. 4D). Apparently there were no explosive eruption products generated during the construction of Cofre de Perote volcano (Fig. 5); however, on the lower western flanks of the volcano, there are some pyroclastic deposits that correspond to the Xaltipan ignimbrite dated at 0.45 Ma (Ferriz and Mahood, 1984). These deposits belong to catastrophic eruptions of Los Humeros Caldera and tend to mantle some parts of the lavas produced during the first stage of Cofre de Perote volcano evolution, as depicted in Figure 4B with a thin line between these two units. However, these pyroclastic deposits lay stratigraphically below lava flows and breccias comprising the third constructional stage of the volcano. Although the variable thickness of this pyroclastic layer is generally unknown, they can be estimated in the range of tens of meters. The final episode of eruptive activity occurred along the outer flanks (not shown) and is related to parasitic cinder cones located northeast of the main Cofre de Perote volcano edifice. The most recent of this late-stage activity corresponds to El Volcancillo, a cinder cone dated at 900 yr B.P. (Siebert and Carrasco-Núñez, 2002).


Fieldwork and Sample Collection

We limited most of our fieldwork to the proximal parts (Fig. 3) of the studied deposits. Therefore, the Xico avalanche was the most thoroughly sampled of the two deposits (Fig. 6). Nonetheless, the Los Pescados debris flow was also sampled at a few locations along Los Pescados River, which is the main eastern-flowing drainage of Cofre de Perote volcano and the northern sector of the Citlaltépetl–Cofre de Perote volcanic range (Fig. 6). For comparative purposes, Figure 6 shows the mapped extent of these and other major debris-avalanche and debris-flow deposits along the main drainage pathway along Los Pescados River and several of the tributary drainages that converge toward it (Carrasco-Núñez et al., 2006). Samples include those collected from the Cofre de Perote volcano summit area, Xico avalanche, Los Pescados debris-flow, and adjacent debris-avalanche and debris-flow deposits from nearby Citlaltépetl–Cofre de Perote volcanic range sources, which were sampled for comparison and to clarify possible confusion between likely volcanic source areas.

A few samples (e.g., CP-0525, CP-0526, and CP-0527; Fig. 6) were collected at locations where other avalanche deposits have been reported, such as one from Las Cumbres volcano (Rodríguez, 2005), which is also exposed near the convergence of Los Pescados debris flow and Xico avalanche. At each location, we retrieved at least 1 kg of mostly matrix material, as our primary focus is on the most dominant minerals in the clay-sized fractions, except for hydrothermally altered rock samples from the Cofre de Perote volcano summit area, which is better exposed (Fig. 7). The Los Pescados debris-flow and Xico avalanche samples were collected from steep, vertical outcrops in order to avoid thick vegetation and soil cover. Despite this, several of our sample sites displayed <50-cm-thick, weathered, and indurated surfaces that were removed in order to expose underlying “fresher” deposit material. In some cases, this weathered surface included a 2- to 3-cm-thick organic layer. In order to distinguish between alteration minerals formed by hydrothermal processes from those formed by chemical weathering processes, we were careful to remove as much soil contamination as possible before sampling. All samples were collected during the dry season in order to avoid the need for extensive drying in the laboratory, which would alter certain clay minerals of interest. The following are field descriptions of the Los Pescados debris-flow and Xico avalanche deposits and their source area. Detailed descriptions of other deposits shown in Figure 7 are provided by Carrasco-Núñez et al. (2006).

Cofre de Perote Volcano Summit Area

The summit area shows spectacular horseshoe-shaped scarps (Fig. 3), which provide the most obvious evidence of major collapse episodes in the past. The shield-like morphology of Cofre de Perote volcano is truncated at the SE flank (Figs. 5 and 7) by these fresh, steep-looking scarps, which show the volcano's inner structure. These scarps expose a sequence of andesitic-dacitic lava flow beds with large irregular areas showing the more permeable and hydrothermally altered rocks (Fig. 7). Several samples of hydrothermally altered rock material (e.g., CP-0515, CP-0519, and CP-0520; Fig. 6) were collected mainly from safely accessible, fractured, and brecciated bedrock areas near the main summit scarp (Fig. 7). Sample CP-0637 was retrieved from one of the most pervasively altered exposed zones (circled area 1, Fig 7).

Los Pescados Debris Flow

Los Pescados debris-flow extends for a distance of ∼60 km with variable thickness up to 25 m. The deposit is massive and includes a mixture of boulders and gravels within a strongly cemented clayey-silty matrix; this matrix cement material was undetermined in the field but is dominated by hydrous silica minerals based on laboratory XRD and infrared spectral measurements, both of which are further discussed in detail. Such minerals are quite common in indurated soils and volcanic deposits exposed on the eastern flanks of Cofre de Perote volcano (e.g., Elsass et al., 2000). Hubbard et al. (2007) were able to estimate a volume of 0.35 km3 for the Los Pescados debris-flow deposit based on deposit thickness, planimetric area, total runout, and geographic information system (GIS)–based, cross-sectional lahar mapping methods.

At the convergence between Río Los Pescados with Río Huitzilapan (Figs. 1 and 6), relatively flat terraces can be seen (“Río Pescados debris flow”; Hubbard et al., 2007). Along its flow path, it shows inverse gradation with an abundant silty-clayey matrix. There are also some jigsaw-shaped blocks present, which are typical of avalanche deposits; however, no hummocks are observed. Lithic fragments inside the deposit are similar to those found at the summit of Cofre de Perote volcano (Fig. 5); in fact, these fragments display aphanitic texture comprised of large crystals of plagioclase with abundant pyroxenes, and they are dominated by andesite and basalt lava flows and breccias. Carrasco-Núñez et al. (2006) proposed that the Los Pescados debris flow was derived from the rapid transformation from a debris avalanche to debris flow, similar to the behavior of the Teteltzingo lahar at Citlaltépetl volcano (Carrasco-Núñez et al., 1993). The age of Los Pescados debris flow has been estimated using 14C as 42 ka (Carrasco-Núñez et al., 2006), which occurred long after cessation of activity from Cofre de Perote volcano.

Xico Avalanche

The most recent deposit associated with the collapsing stage of Cofre de Perote volcano is Xico avalanche, with a mapped area of ∼73 km2 (Fig. 6) and an average thickness of 30 m, which yields a volume of ∼2.19 km3 (Díaz-Castellón, 2009). Xico avalanche age ranges from ∼11–13 ka based on a 14C date derived from wood collected from the deposit (Carrasco-Núñez et al., 2006). Stratigraphically, Xico avalanche is on top of Los Pescados debris flow but only covers the latter at the proximal and medial extents because Los Pescados debris flow has a longer runout. At proximal locations, Xico avalanche displays the distinctive hummocky topography of debris-avalanche features (Fig. 3). Xico avalanche contains megablocks (26–32 m long) and is clast to clast supported, heterolithologic, and displays jigsaw-shaped fracturing in clasts. Fresh lithic fragments are mostly andesitic rocks containing abundant pyroxene and plagioclase phenocrysts, which resemble summit area rocks from the Cofre de Perote volcano third stage (Fig. 5). These same rocks are also found in Los Pescados debris flow, though in lesser abundance. At the town of Xico (Fig. 6), the average measured depth of the deposit was ∼22 m above old, thick basaltic lava flows, where there is no contact (with the Los Pescados debris-flow deposit).

Laboratory X-ray Diffraction and Visible to Short-Wave–Infrared (0.4–2.5 μm) Reflectance Spectroscopy

Samples of hydrothermally altered rocks from the Cofre de Perote volcano summit and matrix material from the two debris-avalanche and debris-flow deposits were analyzed using both X-ray diffraction (XRD) and visible to short-wave–infrared (VIS/IR herein; 0.4–2.5 μm) reflectance spectroscopic methods. In general, both methods provide complementary mineralogical information that is usually but not always in agreement (e.g., Buckingham and Sommer, 1983). For example, Buckingham and Sommer (1983) and Hunt and Ashley (1979) all note that hematite and goethite can occur as rock coatings that can spectrally mask a prominent absorption feature at 0.43 μm (Fig. 8) related to jarosite. As a result, XRD patterns can show prominent X-ray peaks related to jarosite, but weak to nonexistent peaks related to hematite and goethite, despite often greater abundances of the latter two minerals. Buckingham and Sommer (1983) discuss other examples where VIS/IR reflectance spectra identified hydroxyl-bearing clay minerals that were not detected by XRD.

The origin of VIS/IR spectral absorption features related to minerals characteristic of weathered and hydrothermally altered volcanic rocks is reviewed by Hunt (1977), Hunt and Ashley (1979), and papers referenced therein. For example, aluminous clay minerals such as kaolinite and smectite (Fig. 8) display diagnostic absorption features at ∼1.4 μm and 2.2 μm (Fig. 8) with an optional feature ∼1.9 μm, if water-expandable layers are present. Broader absorption features at wavelengths <1.2 μm are characteristics of hematite (not shown), goethite, and jarosite (Fig. 8), while the sulfate minerals, alunite and jarosite, display prominent features at ∼2.17 μm and ∼2.26 μm, respectively (Fig. 8).

For this study, we make use of laboratory-measured VIS/IR spectra to identify and characterize absorption features related to the most abundant (i.e., spectrally dominant) minerals. XRD provides additional information about the bulk mineralogy of each sample, including mineral phases, which are either not abundant enough to be detected using VIS/IR spectra, or masked by surface coatings dominated by other minerals.

Both methods are complementary and have advantages and disadvantages for field and/or laboratory study of debris-avalanche and debris-flow deposits. For example, unlike XRD and other laboratory-based analytical methods, VIS/IR spectral-reflectance measurements do not require elaborate sample preparation methods that can either alter the sample from the way it occurs naturally or require small or specific portions of the sample that may not be representative of the material from which it came (Crowley and Vergo, 1988a). However, VIS/IR spectra generally penetrate only the upper few tens of micrometers of the surface being measured (Buckingham and Sommer, 1983), unlike XRD, which yields volumetric-based mineral abundances. Using one of earliest portable field spectrometers available, Marsh and McKeon (1983) note that dominance or interference of absorption features between mineral components within intimate mixtures (Clark, 1999) sometimes necessitates further XRD analysis, which can only be done in the laboratory. We likewise solve this problem by measuring XRD on the same powdered samples we measured using VIS/IR spectra, although the results of each analytical method were interpreted independently of one another.

For XRD and VIS/IR spectral analysis, all of our samples were mechanically sieved to separate the coarser sand fractions dominated by pyrogenic minerals (i.e., those of igneous or volcanic origin) of little or no interest such as feldspars, quartz, pyroxenes, amphiboles, and unaltered vitric components, from the silt- and clay-size fractions using standard sieves with meshes ranging in diameters from –6 to 4 φ. Further, mineralogical analysis was conducted using powders containing the smallest size fraction (i.e., very fine sand and smaller <4 φ) in order to concentrate minerals formed possibly by either hydrothermal alteration or chemical weathering processes. Some samples of hydrothermally altered rocks from the Cofre de Perote volcano summit area were crushed and/or disaggregated prior to sieving. Notably, the intensity of pyrogenic mineral peaks in XRD data can be used as an inverse proxy measure of alteration intensity, as we show in our results using feldspar peaks.

For XRD, we used a Scintag X-ray diffractometer using CuKa radiation with an intensity of 45 kV measuring a total 2-theta (i.e., 2θ) interval from 2° to 40° @ 0.02 steps per interval. Both the original random powdered sample (crushed <4 φ) and clay-sized (<2 μm) fractions were measured. Following the methods of Moore and Reynolds (1997), oriented clay mounts were prepared by separating the clay-sized fraction left in suspension from the coarser fractions settling in distilled water for ∼4 h unless flocculation occurred. Ethylene glycol saturation and heat treatments were done to test for expandable and mixed-layered clays, chlorite and hydroxy-interlayered clays (Moore and Reynolds, 1997; Meunier, 2007), the latter of which is formed in soil environments (Barnhisel, 1977). Identification of many crystalline, nonclay mineral phases were easily done using automated X-ray peak matching software and Joint Committee on Powder Diffraction Standards (JCPDS) for pure minerals. However, identification of phyllosilicates required reference to published patterns, mostly from Moore and Reynolds (1997) and references therein. Feldspar peaks were identified using our own powdered standards representing solid-solution compositions typical for andesitic and dacitic volcanic rocks exposed throughout the Citlaltépetl–Cofre de Perote volcanic range (e.g., Negendank et al., 1985).

Visible to short-wave infrared reflectance spectra were measured using a field-portable analytical spectral device (ASD) measuring visible to short-wave infrared reflected radiation from 0.4 to 2.5 μm. All sample powders were measured against a dark carbon background, using a quartz lamp as an artificial light source, and brightness values were measured relative to “Spectralon,” a high-reflectivity plastic standard. Spectral absorption features were identified and visually interpreted using spectral library and published data for various minerals, including mixed-layered clays and various grain sizes of the same mineral (Crowley and Vergo, 1988a, 1988b; Clark et al., 1993).

ASTER and Hyperion Remote-Sensing and Spectral Analysis

For this study, we map hydrothermally altered rocks at Cofre de Perote volcano using multispectral Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data acquired 29 March 2002 onboard the Terra satellite and hyperspectral imagery acquired 28 November 2001 by the EO-1 Hyperion sensor. Hyperion provides continuous (196 out of 242 channels) coverage from 0.4 μm to 2.4 μm at spectral resolution of ∼10 nm (Folkman et al., 2001). In particular, ASTER imagery was acquired most optimally under cloud-free conditions during the height of the dry season.

ASTER measures reflected radiation in three bands between 0.52 μm and 0.86 μm (visible and near-infrared [VNIR]) and in six bands from 1.00 μm to 2.43 μm (short-wave infrared [SWIR]), with 15 m and 30 m resolution, respectively (Fujisada, 1995). Level 1B VNIR and SWIR “at-sensor” radiance data were spatially resampled to the same 15 m resolution, coregistered, atmospherically corrected, and calibrated to reflectance using “ACORN” software (ImSpec LLC, 2004).

Although cloud-free ASTER coverage exists over the entire mapped extents of the Xico avalanche and Los Pescados debris-flow deposits (Fig. 6), the medial and distal portions of the deposits between 2000 m and 500 m, respectively, are covered by dense, mixed montane oak forests, which grade into tropical rainforest, open savannas, and tropical deciduous forest toward the coast of Veracruz (Lauer, 1973). Between 2000 m and 3500 m, clouds bank up against the Citlaltépetl–Cofre de Perote volcanic range to produce montane cloud forests (Lauer, 1973), which obscure the volcanic deposits and soils of much of Cofre de Perote volcano from remote-sensing mapping. Therefore, we spatially subsetted our nine-band ASTER reflectance data in order to focus on the uppermost portions of the Cofre de Perote volcano summit cone, especially areas above the ∼3500 m timberline (Lauer, 1973), where grasses gradually thin out and are replaced by bare rock and tephra exposures (Figs. 5 and 9).

Spectral end members representing the three most common (i.e., abundant) classes of altered rocks exposed on the Cofre de Perote volcano summit (Fig. 9) were derived using an automated pixel-purity analysis procedure (Boardman et al., 1995), after masking out vegetated pixels on the lower slopes. Pixels with the strongest spectral absorption features, often diagnostic of individual minerals and mineral mixtures characteristic of hydrothermally altered rocks, were extracted and used as spectral end members based on principal components analysis–based, multidimensional image transformations described in detail elsewhere (e.g., Green et al., 1988; Boardman, 1993; Boardman et al., 1995). These spectral end members were then used as reference image spectra for subsequent match-filter classification (Harsanyi and Chang, 1994) of the entire ASTER scene, the results of which are discussed in the subsequent section.

Because of Hyperion's narrow swath coverage (∼7.7 km wide), limited spatial resolution (30 m/pixel), and lower signal-to-noise (1/6 that of ASTER SWIR; Hubbard et al., 2003), as well as the persistent cloud cover over the Cofre de Perote volcano summit area and limited exposure of the most intensely altered rocks near the summit (Fig. 7), the Hyperion data were used for spectral analysis instead of for mapping purposes. However, the Hyperion data were also calibrated using the same ACORN software (ImSpec LLC, 2004) used to calibrate the ASTER data.


Hydrothermal Alteration Mapping of the Cofre de Perote Volcano Summit Area

Figure 9 shows ASTER spectral matched-filtering results displaying the distribution of three most dominant classes of altered rocks on the Cofre de Perote volcano summit and mapped as yellow, orange, and red, respectively: (1) areas dominated by smectite mixtures and characterized by strong smectite absorption features in both the ASTER and Hyperion data (e.g., circled yellow areas 1 and 3, Figs. 7, 9, and 10); (2) areas characterized by moderate to weak ferric-iron slopes between ASTER bands 2 and 1 and weaker Al-OH clay absorption features at 2.2 μm, which correspond to ASTER band 6 (e.g., area 3 spectral plots, Fig. 10); and (3) areas characterized by strong ferric-iron features between ASTER bands 2 and 1, but no resolvable Al-OH clay absorption features. The third and latter class of iron-oxide (e.g., hematite) altered rocks (red-colored areas, Fig. 9) could be related to rapid in situ degassing of edifice-building lava flows and tephra (e.g., Crowley and Zimbelman, 1997), or slower chemical weathering of magnetite and other iron-bearing phases within these rocks (e.g., Buckingham and Sommer, 1983) as opposed to hydrothermal alteration. In the case of Cofre de Perote volcano, these areas are more exposed and more extensively mapped on the western flanks because it is more arid and has less vegetation cover than the eastern flanks (Fig. 9).

Circled areas 1–3 (Fig. 9) are among the most intensely altered areas of the Cofre de Perote volcano summit; these areas are easily visible on the ground as colored zones distinct from the surrounding bedrock areas (corresponding circled areas 1–3, Fig. 7). Although small in size (<30 m for intensely altered bedrock areas), these areas plus their downslope talus aprons were well resolved spatially and spectrally using ASTER (Fig. 9). There are several reasons for this, including: ideal off-nadir viewing conditions (i.e., –8.5° pointing angle), enhanced SWIR signal-to-noise, effective calibration, 15-m resolution VNIR bands that are sensitive to ferric-iron minerals in altered areas, and talus derived from the alteration zones that exaggerates their mapped extents. Notably, these same areas mark permeable ash and tephra layers, which are more susceptible to hydrothermal alteration (Watters et al., 2000).

Circled area 4 (Fig. 9) is located off the photograph shown in Figure 7, but it is easily accessible via the road leading to the summit area, and radio facilities (Fig. 7). Pixels defining the road were mapped because nearby rocks, including altered rocks, are used as localized sources of pavement gravel material. Similarly, the most proximal exposures of the Xico avalanche deposit (sample sites CP-197 and CP-0501; Fig. 9) were mapped well along the roadside despite the dense vegetation cover on the eastern flanks for this same reason. At these two sample locations, it is quite possible that the road mapped using ASTER (Fig. 9), traces the extent of outcropping exposure of smectite-rich matrix from both the Xico avalanche deposit and an underlying pre–Xico debris-avalanche deposit discussed in a later section. This smectite clay-rich material could be a considerable hazard for driving along mapped portions of this road during the rainy wet season.

We collected and analyzed four representative samples from the Cofre de Perote volcano summit area for ground truth (Figs. 6, 7, and 9). Figure 11 shows XRD and VIS/IR spectral analysis results for the four Cofre de Perote volcano summit samples. Sample CP-0637 represents the most intensely (i.e., pervasive), hydrothermally altered rocks exposed at the Cofre de Perote volcano summit. In this case, both the XRD (Fig. 11A) and VIS/IR spectra (Fig. 11B) agree and display X-ray peaks and visible-infrared absorption features related to jarosite and a kaolinite-smectite mixture, though smectite appears to be slightly more abundant than kaolinite. The VIS/IR reflectance spectrum (Fig. 11B) lacks the secondary jarosite absorption feature at 2.26 μm, perhaps due to masking by abundant clays (e.g., Hunt and Ashley, 1979). However, it does display the prominent 0.43 μm ferric-iron charge transfer feature diagnostic of jarosite (Figs. 8 and 11B), and 0.66 μm absorption, which is much stronger for goethite than it is for hematite. Notably, hematite and/or goethite were not abundant enough to be detected in the XRD (Fig. 11A), even though minute amounts (i.e., a few percent) of these minerals can be readily detected using VIS/IR spectroscopy (Buckingham and Sommer, 1983; Clark, 1999).

Sample CP-0519 (Figs. 11E and 11F) displays a similar mineralogy to sample CP-0637 collected nearby (both within circled area 1; Fig. 9), but its XRD pattern (Fig. 11E) shows that it contains alunite in addition to jarosite. Alunite displays a prominent absorption feature at 2.165 μm, which overlaps with part of the kaolinite doublet absorption at 2.2 μm (Fig. 8), and has a secondary absorption feature at 2.325 μm. Sample CP-0519 lacks these diagnostic alunite absorption features (Fig. 11F) due to masking by more abundant kaolinite (e.g., Hunt and Ashley, 1979), which displays diagnostic doublet absorption features at both 1.4 μm and 2.2 μm. A large interlayer water absorption feature is prominent at 1.9 μm (Fig. 11F), for which XRD patterns confirm the presence of expandable smectite layers.

Comparison of samples CP-0637 and CP-0519 with laboratory spectra (Figs. 11B and 11F, respectively), ASTER and Hyperion image spectra from the same location (Fig. 10A), and spectral library spectra of smectite, kaolinite, and various mixtures of the two minerals (Fig. 12) shows interesting results. Despite random and coherent noise, the 196-band Hyperion spectrum (middle, Fig. 10A) resolves the smectite singlet absorption well (e.g., Figs. 8 and 12), even after it is convolved to the nine-point band passes of ASTER for comparative purposes. However, seasonal differences between the imagery are noticeable when comparing the ASTER image spectrum (top, Fig. 10A) with the Hyperion convolved to ASTER image spectrum (bottom, Fig. 10A), the latter of which underwent noise reduction in the convolution process. The ASTER spectrum shows a slight chlorophyll absorption in band 2 (0.66 μm), which is related to subalpine grasses that grow at elevations as high as 4000 m (Barois et al., 1998). This elevation contour corresponds to the level somewhat below the main avalanche-scarp cliff face shown in Figure 7. The corresponding Hyperion spectrum shows a slight absorption feature near 2.3 μm, which is either due to dry (or senescent) grass or noise. None of these vegetation-related features are found in the sample spectra (Figs. 11B and 11F).

The ASTER and Hyperion image spectra (Fig. 10B) of circled area 3 (Figs. 7 and 9) show similar spectral features in both data sets, though not quite as well resolved as in the case of area 1. This is perhaps due to the smaller size of this feature and oversampling of its aerial extent within the 30-m pixels of Hyperion and ASTER SWIR bands. Despite this problem, the ASTER spectrum (top, Fig. 10B) displays the highest slope between the 15-m resolution bands 2 and 1, which is indicative of high ferric-iron content at this location. Because of its higher signal-to-noise, ASTER can better separate aerial mixtures abundant in kaolinite from aerial mixtures abundant in smectite, using fewer bands than Hyperion (e.g., Hubbard et al., 2003). In this case, image spectra derived from both data sets (Fig. 10) lack clear evidence of kaolinite features, and are all dominated by smectite absorption features. This suggests that smectite is perhaps the most abundant alteration mineral at the scale of ASTER and Hyperion pixels for these most intensely altered areas (yellow colored, Fig. 9). Also, our samples from these areas contain smectite intimately mixed with other minerals, which may not be fully representative of the larger scale (i.e., 30 m), remote-sensing pixel areas from which they were derived.

Zimbelman (1996) ranks the intensity of hydrothermally altered rocks at Mount Rainier and other Cascade volcanoes as ranging from incipient, where volumetrically smaller portions of the original rock (e.g., glassy groundmass) are altered; to partial, where major (e.g., tens of percent) portions of the original rock are altered; to pervasive, where the entire volume of original rock, including crystalline phases, are replaced by secondary clay and/or sulfate minerals. He also notes several modes of occurrence, which vary from replacement (the most common), to fracture-filling (e.g., vein and stock structures common in ore deposits) and encrustation (i.e., coating) types.

The XRD patterns of area 4 samples CP-0515 (Fig. 11C) and CP-0520 (not shown) are dominated by plagioclase feldspar, suggesting incipient alteration, although their VIS/IR spectral-reflectance patterns (Fig. 11D) display broad singlet absorptions at 2.2 μm indicative of smectite. Visible to short-wave infrared reflectance spectroscopy is more sensitive to clay mineral coatings than XRD (Buckingham and Sommer, 1983), especially when they are present in low concentrations such as in incipiently altered rocks represented by samples CP-0515 and CP-0520. However, the smectite absorption features displayed by samples CP-0515 and CP-0520 are much broader than those displayed by a pure spectral library sample (Fig. 8). This broadening could be due to mixing with amorphous hydrous silica minerals such as Opal-A (Hunt and Ashley, 1979), allophane (Cooper and Mustard, 1999), or grain-size variations related to particle-packing differences on the surface of smectite-bearing coating materials (Cooper and Mustard, 1999).

Samples CP-0515 and CP-0520 (circled area 4, Fig. 9) are perhaps the most representative samples of the second and most widespread class of altered rocks mapped using ASTER (orange-colored areas, Fig. 9). These areas are characterized by moderate to weak ferric-iron slopes between ASTER bands 2 and 1, and relatively weaker (than yellow-colored areas, Fig. 9) yet still resolved Al-OH absorption features at 2.2 μm (ASTER band 6). We interpret the distribution pattern for this class of altered rock as coinciding with coatings, encrustations, and fillings along the heavily fractured bedrock areas of the Cofre de Perote volcano summit (Fig. 5), which can best be characterized by incipient to partial in alteration intensity. In contrast, yellow-colored areas (Fig. 9) represented by samples CP-0637 and CP-0519 contain much less feldspar and more abundant clay, sulfate, and ferric-iron minerals, which are most indicative of pervasive alteration intensity.

Additional areas mapped as the second class of altered rocks (orange areas, Fig. 9) are found on the lower southwestern flanks of Cofre de Perote volcano. This includes the town of Los Altos (Fig. 9), which sits on top of alluvial deposits and soils near ∼3000 m (Fig. 5). Soil profiles (i.e., pedons) (e.g., circled area 5, Fig. 9) and chemical weathering rates in this area were characterized by Dubroeucq et al. (1998), who report abundant 10-angstrom (Å) halloysite at depth (<230 cm), which becomes progressively dehydrated as 7 Å halloysite toward the surface. Numerous other pedons characterized at this (3100 m) and lower elevations around the eastern and western flanks of Cofre de Perote volcano show halloysite to be the dominant chemical weathering mineral in andosols derived from volcanic deposits in the area (Barois et al., 1998; Dubroeucq et al., 1998; Elsass et al., 2000; Dubroeucq et al., 2002).

Because kaolinite- and smectite-intimate mixtures often resemble halloysite, even when using hyperspectral data (e.g., Crowley and Vergo, 1988a; Hauff et al., 1990; Fig. 12), it is possible and quite likely that during spectral classification of our nine-band ASTER data, halloysite-bearing soils and alluvial deposits exposed below timberline were confused with altered rocks containing smectite and kaolinite mixtures above timberline (orange areas, Fig. 9). Notably, the town of Los Altos was mapped well in the imagery because its roads and building roofs are constructed using local sources of clay-rich building materials, such as soils and weathered volcanic deposits.

Clay Mineralogy of the Los Pescados Debris‑Flow Deposit

Granulometric analyses on selected samples of the Los Pescados debris-flow were done to determine fine fraction distribution. Sample CP-0514 was initially analyzed for granulometric analysis, though further samples were needed. For this reason, we collected additional samples (Fig 6). Analysis show that the deposit varies in grain size from 86% to 32% of gravel clasts, 58% to 13% for sand clasts, 3.3% to 0.5% for silt clasts, and 0.4 to 0 for clay particles (Table 1). Grain size of particles below 4 φ indicates a low fine-grained content, whereas clay content is below 1%, confirming that it is noncohesive in origin. Comparably, cohesive debris flows and lahars generally contain between 3% and 5% clay and have higher silt contents (Crandell, 1971; Scott et al., 1995).

Figure 13 shows VIS/IR laboratory reflectance spectra measured for the silt plus clay fractions of Los Pescados debris-flow samples CP-184, CP-0514, and LP-01. All three samples display similar spectral absorption features, which are not conclusive due to intimate mixing between clays and other hydrous minerals. For example, they all display weak doublet absorption features at 1.4 μm, which are diagnostic of either halloysite or kaolinite. Also, they display interlayer water absorption features at 1.9 μm, which are diagnostic of halloysite, smectite, or other expandable clay mineral. The most diagnostic absorption features for di-octahedral aluminous clays are at or near 2.2 μm, which, in the case of Los Pescados debris-flow samples, are broad due to either mixing between clay and/or hydrous silica minerals, or perhaps grain-size and particle-packing effects (Cooper and Mustard, 1999).

For these samples, XRD provided more conclusive mineralogy results for the three Los Pescados debris-flow samples (Fig. 14). All three samples contain dehydrated and/or hydrated forms of halloysite, and display large peaks at 22.18° 2θ due to cristobalite peaks superimposed on one of the smaller but major peaks of plagioclase feldspar (Fig. 14). Cristobalite and opal are common components of soils and weathered volcanic rocks (Wilding et al., 1977) and especially at Cofre de Perote volcano (Barois et al., 1998; Dubroeucq et al., 1998; Elsass et al., 2000; Dubroeucq et al., 2002).

Clay Mineralogy of the Xico Avalanche and Similar Deposits

Xico avalanche samples generally show better agreement (i.e., less confusion) between VIS/IR spectral-reflectance results (Fig. 15) and XRD results (Fig. 16) than the Los Pescados debris-flow samples (see Fig. 6 for sample locations). This is due in part to much stronger absorption features, and hence greater abundances of clay minerals in the Xico avalanche samples than in the Los Pescados debris-flow. To facilitate discussion, VIS/IR reflectance spectra (Fig. 15) have been grouped and/or stacked together to highlight samples containing similar clay mineral compositions. For example, samples CP-0524, CP-0512, and CP-165 (Fig. 15A) are each dominated by strong kaolinite doublet absorption features at 1.4 μm and the more diagnostic 2.2 μm wavelength positions. Corresponding XRD patterns (Figs. 16A, 16B, and 16C) confirm the presence of strong 7 Å kaolinite peaks in the d001 position (i.e., ∼12° 2θ), with good preferred orientation in the d002 and d003 positions (i.e., ∼20° and ∼35° 2θ, respectively). Sample CP-165 also displays a prominent water absorption feature at 1.9 μm (Fig. 15A). The corresponding XRD pattern indicates that this is due to a hydroxy-interlayered smectite (HIM; Fig. 16C), which yields a 14 Å peak that subsequently collapses to 10 Å after heating (Meunier, 2007). Aluminum (and/or iron) hydroxy-interlayering suggests pedogenic modification of smectite, perhaps formed originally as a result of hydrothermal alteration. Neither kaolinite nor smectite occurs abundantly in soils in and around Cofre de Perote volcano (Barois et al., 1998; Dubroeucq et al., 1998; Elsass et al., 2000; Dubroeucq et al., 2002).

Notably, the XRD patterns for samples CP-0524 and CP-0512 (Figs. 16A and 16B, respectively) show additional strong peaks related to hematite and cristobalite, which are either weak or not apparent in their corresponding VIS/IR spectral-reflectance patterns (Fig. 15A). This can be explained by the concentration of some minerals on the surface of the sample for which VIS/IR reflectance measurements are most sensitive, while others may be concentrated throughout the bulk volume of the sample for which XRD is most sensitive (e.g., Buckingham and Sommer, 1983).

Xico avalanche samples CP-0523 and CP-0511 (Fig. 15B) are both dominated by strong smectite-related absorption features at 1.4, 1.9, and 2.2 μm. The corresponding XRD pattern for CP-0511 (Fig. 16D) confirms that expandable smectite is the most abundant mineral in this sample; however, it also contains smaller amounts of gypsum and jarosite, both of which are the clearest evidence of hydrothermal alteration origin yet. Similar to XRD and VIS/IR reflectance patterns of Cofre de Perote volcano summit samples CP-0515 and CP-0520 (Figs. 11C and 11D, respectively), Xico avalanche sample CP-0523 yields a VIS/IR reflectance pattern dominated by smectite (Fig.16B), but an XRD pattern (Fig. 16E) showing no evidence of it. This may be the result of grain size, packing, and coating effects (e.g., Buckingham and Sommer, 1983; Cooper and Mustard, 1999) discussed previously with respect to other samples showing disagreement between XRD and VIS/IR spectral data. However, unlike Cofre de Perote volcano summit samples CP-0515 and CP-0520, Xico avalanche sample CP-0523 does display relatively lower intensity plagioclase feldspar peaks, a strong cristobalite (i.e., hydrous opal-C?) peak, and weak hematite peaks (Fig. 16E). The latter two minerals agree with the corresponding VIS/IR spectral-reflectance pattern (Fig. 15B).

Xico avalanche samples CP-0521 and CP-197 (Fig. 15B) and CP-155 and AX-1 (Fig. 15C) all exhibit spectral absorption features similar to each other and the three Los Pescados debris-flow samples (Fig. 13). However, the XRD patterns of CP-0521 and CP-197 (Figs. 16F and 16G, respectively) are dominated by hydrated and dehydrated forms of halloysite, while the XRD patterns of CP-155 and AX‑1A-X1 (Figs. 16H and 16I, respectively) are dominated by expandable smectite mixed with either kaolinite in the case of AX-1 (Fig. 16I) or halloysite in the case of CP-155 (Fig. 16H). As discussed in a prior section, intimate mixtures of smectite plus kaolinite (and/or other types of kaolin minerals) are difficult to distinguish from halloysite using VIS/IR reflectance spectra (Fig. 12). Sample AX-1 is particularly interesting because it shows evidence of mixed-layered illite-smectite (MIS), in addition to containing kaolinite (Fig. 16I). This particular clay-mineral assemblage (MIS plus kaolinite) has been documented in hydrothermal alteration environments but is not common in near-surface (i.e., shallow) soil-forming environments, even if K-rich sedimentary and igneous rock substrates are involved (Moore and Reynolds, 1997, p. 172–183). Also, samples CP-155 and A-X1 both contain jarosite, which is certainly of hydrothermal origin (Figs. 16H and 16I, respectively), while samples CP-0521 and CP-197 do not (Figs. 16F and 16G, respectively).

Notably, CP-197 is the proximal-most sample of the Xico avalanche collected (Fig. 6). It was collected along a road leading up to the ∼3000 masl elevation mark along the eastern flanks of Cofre de Perote volcano (sample CP-197; Fig. 9). An older debris-avalanche deposit lying stratigraphically below the Xico avalanche was sampled along this same road (sample CP-0501; Fig. 9). However, preliminary XRD and VIS/IR spectral analysis results (not shown) both show a mixture of smectite and halloysite as the dominant clay minerals in this older (pre–Xico avalanche) debris-avalanche sample, the age and extent of which has not yet been studied or mapped (see Figs. 6 and 9 for location).

Xico avalanche samples CP-0510 and CP-0501 both illustrate examples of VIS/IR spectra (Figs. 15C and 15D, respectively) dominated by strong smectite absorption features at 1.4 and 2.2 μm and/or weak halloysite-related absorption features at these same wavelengths. The XRD patterns of both samples (Figs. 16J and 16K, respectively) confirm that smectite is indeed the most dominant clay mineral in both samples. Also, both samples show XRD evidence of hydrated and/or dehydrated forms of halloysite, which are most likely the result of postemplacement chemical weathering. The XRD patterns show that both samples contain jarosite (Figs. 16J and 16K), with sample CP-0510 exhibiting the most intense jarosite peaks, and therefore the most abundant amounts of jarosite evident in all the Xico avalanche samples collected. Jarosite abundance in sample CP-0510 is clearly enough to produce a prominent 0.43 μm ferric-iron charge transfer feature diagnostic of jarosite in the corresponding VIS/IR reflectance spectra for this sample (Fig. 15C). However, other diagnostic jarosite absorption features are not well resolved due to masking by more abundant clay minerals.

Xico avalanche samples CP-0524 and CP-0527 are two out of 12 Xico avalanche samples that do not contain detectable amounts of feldspar, quartz, and/or pyroxene within their <4 φ sized fractions (Figs. 16A and 16L, respectively). Notably, quartz is the most weathering resistant of three pyrogenic contaminant minerals contained in our samples. Feldspar and pyroxene minerals are less weathering resistant than quartz. However, samples CP-0524 and CP-0527 differ from one another in that the former is dominated by abundant kaolinite (Fig. 16A), while the latter is dominated by abundant halloysite (Fig. 16L). In particular, the XRD pattern (Fig. 16L) and VIS/IR reflectance pattern (Fig. 15D) of sample CP-0527 both show X-ray peaks and spectral absorption features related to the presence of halloysite and gibbsite. Gibbsite and aluminous hydroxy-interlayered 2:1 layered clays are both strong indicators of intense chemical weathering of volcanic deposits, and are also known to coexist in volcanic soils under certain conditions (e.g., Ndayiragije and Delvaux, 2003).


A summary of XRD results is provided in Table 2 for all samples analyzed in this study, including additional random powder, oriented, and specially treated patterns not shown in Figures 11, 14, and 16. Table 3 summarizes the dominant minerals found in each sample based on VIS/IR spectral analysis results. Results from XRD and VIS/IR spectral analysis of samples of debris-avalanche/flow deposits not related to Cofre de Perote volcano (see Fig. 6 for sample locations) are also shown. The origins of some of these deposits are uncertain (“UN” labeled samples, Tables 2 and 3) because of their proximity to more than one deposit and lack of distinguishing field characteristics, such as lack of hornblende, a key indicator of volcaniclastic rocks from Las Cumbres volcano (Scuderi et al., 2001; Rodríguez, 2005; Carrasco-Núñez et al., 2006). A more detailed description and characterization of possible debris-avalanche and debris-flow source areas for all deposits exposed in the Huitzilapan and Pescados river valleys are summarized elsewhere (e.g., Carrasco-Núñez et al., 2006).

The minerals summarized in Tables 2 and 3 can best be grouped into five classes: (1) pyrogenic (i.e., mostly derived from porphyritic volcanic rocks) contaminant minerals such as quartz, feldspar, pyroxene, and amphibole; (2) sulfate-bearing minerals such as alunite and jarosite, the presence of which provide the strongest evidence of hydrothermal alteration origin; (3) single- and mixed-layered clay minerals such as smectite, kaolinite and illite, and gypsum, which are not common in the soils and weathered volcanic deposits exposed throughout the Citlaltépetl–Cofre de Perote volcanic range and are therefore most likely of hydrothermal alteration origin; (4) minerals such as hematite, goethite, cristobalite, and other types of hydrous silica, which can be either pyrogenic, hydrothermal alteration, or chemical weathering related in origin; and (5) minerals that are clearly of pedogenic (i.e., soil-forming) and chemical-weathering origin such as halloysite, gibbsite, and hydroxy-interlayered clay minerals. For example, alunite was found in only one Cofre de Perote volcano summit area sample (sample CP-0519, Table 2), though not abundant enough to be detected using VIS/IR reflectance spectra (Table 3). Jarosite was found in several Cofre de Perote volcano summit area and Xico avalanche samples (Table 2), two of which contained sufficient amounts to yield diagnostic spectral features suitable to identify them in the 0.4 to 1.0 μm wavelength range (samples CP-0637 and CP-0510, Table 3). Gypsum, though formed by brine concentration in arid evaporate-playa environments (e.g., Eugster and Hardie, 1978), suggests hydrothermal alteration origin for samples collected from the eastern flanks of Cofre de Perote volcano—in contrast to the arid basin of Serdan Oriental on the western side of Cofre de Perote volcano, where such gypsum-bearing saline lakes are numerous.

The matrix of the Los Pescados debris-flow deposit is dominated by halloysite and cristobalite (Tables 2 and 3), which are the two most abundant mineral phases in indurated volcanic soils on the eastern flanks of Cofre de Perote volcano (Elsass et al., 2000, and references therein). Unlike the Xico avalanche, minerals in the matrix of the Los Pescados debris flow appear to be entirely of pedogenic origin, with little or no minerals formed evidently as a result of hydrothermal alteration. Based on this evidence and grain-size analysis data, the Los Pescados debris flow began as a noncohesive or “clay-poor” lahar (Vallance, 2000). The ∼42 ka age of the Los Pescados debris flow corresponds to a period in time in which the ancestral summit of Cofre de Perote volcano was most likely covered by glaciers, as evidenced by glacial cirques on the western flanks of the volcano (Fig. 3). Another possible source of water could have been land-falling tropical cyclones like those that often affect the Gulf Coast of Mexico and Central America today.

We propose two possible sources for the halloysite and cristobalite in the Los Pescados debris-flow deposit. One possibility is that the halloysite and cristobalite formed in situ through chemical weathering after the deposit was emplaced. However, if this was the case, we would see the following evidence in the field: (1) numerous clasts with visible weathering rinds, and (2) uniformly distributed induration of the deposit with weathering profiles as deep as 1–3 m beyond the exposed surface of the deposit (e.g., Elsass et al., 2000). In sampling the deposit, we were careful to avoid the most indurated exposures and collected matrix material beyond the first few meters at each outcrop. Also, larger clasts and blocks of andesite within the deposit, including a few we collected along with our matrix samples, contained no visible weathering rinds. We favor an alternative origin for the halloysite and cristobalite, such that the Los Pescados debris flow bulked (Scott et al., 1995) with clay-rich soils and alluvium prior to being emplaced. This is evidenced by numerous rounded clasts within the Los Pescados debris-flow deposit, indicative of stream transport long distances from their source.

Based on remote-sensing mapping and XRD and VIS/IR reflectance analysis of ground-truth samples, smectite appears to be the most abundant and dominant mineral throughout hydrothermally altered areas exposed high on the Cofre de Perote volcano summit. Similarly, smectite clays such as montmorillinite-beidellite were also found to be dominant in the failure scarp and resulting debris-avalanche deposit at Casita volcano, Nicaragua (Opfergelt et al., 2006). However, in the case of Cofre de Perote volcano, much of this smectite is either mixed with kaolinite and other minerals (e.g., alunite, jarosite, and perhaps hydrous silica) in the most intensely altered ash and tephra layers, or occurs as coating and filling material in the surrounding fractured and partially altered bedrock areas. These same minerals (except for alunite) were also found in Xico avalanche samples (Table 2). Compared to other debris-avalanche and debris-flow deposits we sampled, the VIS/IR spectra of most Xico avalanche samples are dominated by either kaolinite or smectite absorption features, although two samples contain a mixture of the two minerals that could otherwise be confused with halloysite without supporting XRD (Table 3). However, the Xico avalanche deposit also contains halloysite, gibbsite, and hydroxy-interlayered clay minerals, which could only have resulted from chemical weathering after its emplacement.

In contrast to the Los Pescados debris-flow deposit, the Xico avalanche deposit preserves the best mineralogical evidence of a hydrothermal alteration origin of all the deposits we sampled (Table 2). This evidence, together with the large debris-avalanche scarp on the modern summit (Figs. 3 and 7), and the timing of the collapse event (∼11–13 ka), suggests that hydrothermal alteration at Cofre de Perote volcano may have been related to the late Pleistocene glacial retreat that occurred throughout the Citlaltépetl–Cofre de Perote volcanic range, and is speculated to have been a major factor in the emplacement of the 16.1 ka Teteltzingo lahar deposit from Citlaltépetl (Carrasco-Núñez et al., 1993, 2006). Although the timing between the Xico avalanche collapse at Cofre de Perote volcano and the Teteltzingo collapse at Citlaltépetl differs by 3000–6000 years, the existence of an older smectite-rich debris-avalanche deposit near proximal outcrops of the Xico avalanche (PXA sample CP-0501, Tables 2 and 3) suggests that the age of hydrothermal alteration at Cofre de Perote volcano is between 11 (13) and 16 ka.

For hazard-assessment purposes and better context, it is worth comparing the extent of hydrothermally altered rocks mapped at Cofre de Perote volcano using ASTER (Fig. 9) with those mapped at volcanoes in the Cascades, which have also produced large debris avalanches and debris flows in the past (e.g., Crowley et al., 2003). For example, pervasively altered areas of the Cofre de Perote volcano summit scarp mapped using ASTER (yellow-colored pixels; Fig. 9) are only 20%–30% the size, extent, and perhaps volume of altered rocks exposed at Mount Shasta. In the case of the latter, Crowley et al. (2003) estimate volumes of hydrothermally altered rocks at the source areas of major stream drainages at Mount Shasta ranging from 1.2 to 1.8 × 106 m3, after excluding debris on talus slopes. This smaller volume of hydrothermally altered rocks at Cofre de Perote volcano (i.e., <0.5 × 106 m3) would suggest a lesser degree of potential risk from debris-avalanche and debris flow inundation today for populated areas downstream than existed during the time of emplacement of the Los Pescados debris flow and Xico avalanche. However, additional studies using more advanced geophysical methods (e.g., Finn et al., 2001, 2007) are needed to resolve the full extent of both water-saturated and hydrothermally altered rocks buried at depth. Also, future hazards from noncohesive debris flows such as the Los Pescados debris flow cannot be ruled out, assuming the availability of a large enough water source (e.g., an unusually large snowpack and/or a stalled tropical cyclone) and/or additional triggering mechanisms discussed in related work sources (e.g., Díaz-Castellón, 2009; Díaz-Castellón et al., 2009; Carrasco-Núñez et al., 2010).


X-ray diffraction and VIS/IR reflectance spectroscopy provided complementary mineralogical information for studying debris-avalanche and debris flow deposits from Cofre de Perote volcano, although both methods have their unique advantages and disadvantages. Generally, minerals resulting from hydrothermal alteration, such as those contained in our Xico avalanche samples, tend be more easily identifiable using VIS/IR reflectance because they occur in greater abundance and yield stronger, more diagnostic spectral absorption features than weathered rocks and soils. Field spectroscopic measurements can be used to distinguish the weathered, indurated surfaces of outcrops from fresher rock and matrix material that may have been derived from hydrothermal alteration processes. Likewise, these same measurements together with remote-sensing data covering larger areas can be used to map soil mineralogy for comparison with debris-flow deposits derived from the bulking of such materials (e.g., Los Pescados debris flow). Also, the degree of postemplacement weathering and induration of debris-avalanche and debris-flow deposits can be assessed using field spectrometer measurements.

However, aside from portability and battery-life issues when used out in the field, VIS/IR spectroscopy has other limitations when used in either the field or laboratory. For example, intimate mixtures of smectite and kaolinite can resemble halloysite and can require laboratory XRD for positive identification. Spectral measurements of clay and ferric-iron mineral coatings can lead to exaggerated interpretations about their actual abundances, whether within the volume of a sample, or within the surface area covered by remote-sensing pixels. Remote-sensing mapping can be further limited by spatial-, spectral-, and radiometric-resolution issues, cloud cover, and the timing of the plowing and planting season in cases where such data are used for soil-mineral mapping (e.g., Hubbard et al., 2002). Despite these limitations, ASTER data validated by Hyperion spectra were successfully used to map small, pervasively altered areas on the Cofre de Perote volcano summit, and corresponding clay-rich debris-avalanche/flow deposits outcropping along the roads. This demonstrates the utility of the field, laboratory, and remote-sensing methods presented in this paper for the study of debris-avalanche and debris-flow deposits of uncertain origin at Cofre de Perote volcano and other volcanoes around the world.

Special thanks to Consejo Nacional de Ciencia y Tecnología (CONACYT) project no. 44549-F; partial support was provided by “Prospectiva de un ambiente para la formación professional”–Universidad Nacional Autónoma de México (PAPIIT-UNAM) grant # IN106810, National Aeronautics and Space Administration (NASA) grant NAG-57579 and NASA contract S-10224-X with the U.S. Geological Survey. We also thank the State University of New York at Buffalo for allowing us to use their installations, and Dr. Michael F. Sheridan for his aid on XRD analysis. James Crowley, Jeff Wynn, and two anonymous reviewers provided helpful comments that allowed us to improve the manuscript. Gabriel Origel provided additional comments on our ASD analysis. Laboratory technician support at Centro de Geociencias was provided by M.Sc. Sara Solis Valdez. Lic. Teresa Soledad Medina provided an essential review of documented gray literature references. Logistic support was provided by Centro de Geociencias (UNAM). This work is dedicated in loving memory of my father, Rodolfo Senior.

These authors contributed equally to this work.