1 Introduction

The magnitude of explosive eruptions and hazards associated with a volcano is directly related to the Volcanic Explosive Index (VEI; Newhall and Self 1982), a scale that classifies eruptions from 0 to 8 based on their volume of magma erupted and eruption column height. These physical parameters are obtained through the construction of isopach and isopleth maps (e.g., Carey and Sparks 1986; Pyle 1989), which are the result of a detailed stratigraphic survey of a tephra-fall layer; this helps understand the eruptive dynamics of highly explosive eruptions (VEI ≥ 4) (e.g., Avellán et al. 2014). These violently explosive eruptions are known as Plinian-type eruptions (e.g., Cioni et al. 2000) and are closely linked to highly viscous and volatile-rich felsic magmas. Examples of Plinian eruptions around the world include the 12 kyrs Apoyeque eruption (Avellán et al. 2014), the 10.5 kyrs Nevado de Toluca eruption (Arce et al. 2003), the 79 Mount Vesuvius eruption (Sigurdsson et al. 1985), the ∼AD 1315 Tarawera eruption (Nairn et al. 2004), and the 1980 Mount Saint Helens eruption (Carey et al. 1990).

Manizales is a medium-sized city in Colombia (Fig. 1A and B) where more than 400,000 inhabitants (www.dane.gov.co) live above more than a dozen tephra layers. These layers, which have individual thicknesses up to 80 cm, have been mainly associated with Plinian-type activity of the Cerro Bravo dacitic volcano, which is located ~ 25 km east of the city (Fig. 2A) (Herd 1982; CHEC 1983; Thouret et al. 1985; Calvache et al. 1987; Lescinsky 1990; Diaz 1991; Monsalve 1991; Monsalve and Núñez 1992; Gil 2001). This volcano is part of the San Diego–Cerro Machín Volcano-Tectonic Province (SCVTP, Martínez et al. 2014; Murcia et al. 2019), a 140 km-long volcanic chain related to the current subduction of the Nazca Plate underneath the South American Plate (e.g., Bourdon et al. 2003). In this chain, however, there are several other nearby composite volcanoes that could potentially affect Manizales city (Figs. 1C, 2BC). Despite this, there are few detailed studies focused on defining the pyroclastic fall hazard for the city, perhaps because nowadays hazard evaluation is strongly focused on the active Nevado del Ruiz volcano (e.g., Londoño 2016; Gómez-Arango et al. 2018).

Fig. 1
figure 1

Location map of the study area. A Manizales city in the context of the country (Colombia). B Panoramic view of Manizales city. C Digital elevation model of the study region. Note the stratigraphic sites. D The Manizales tephra sequence

Full size image
Fig. 2
figure 2

Panoramic view of volcanoes near Manizales. A Cerro Bravo volcano. B Plazuela volcano. C Nevado del Ruiz volcano

Full size image

This study presents the results of an integrative stratigraphic survey of four of the youngest, thickest, and most representative tephra layers in the urban area of Manizales (Fig. 1D). The study interprets the mapping and physical descriptions of the deposits and evaluates radiocarbon dating of paleosols and calculation of physical parameters. The results lead to the definition of the distribution of the tephra falls, the ages of the events, the source of the volcanic eruptions, the volume of ejecta magma, the columns height and eruptive styles, and the emission rates and eruptions durations. This information was used to give insights into the volcanic hazard for Manizales city and for other smaller urban centers around it.

2 Methods and analytical techniques

2.1 Distribution, tephra thickness, and fragment sizes

For this study, four well-preserved tephra layers were analyzed at 53 stratigraphic sites in the region, with Manizales city as the type location. These units were chosen because they are the thickest layers, and therefore excellent stratigraphic markers (Fig. 1D). Then, the layers were followed both eastward to the Central Cordillera where the active volcanoes are located, and westward up to 32 km away from the Cerro Bravo volcano. Near the Central Cordillera, however, it was not possible to distinguish the four tephra layers among the many pyroclastic fall deposits (more than a dozen) that have originated in the area. The similar mineralogical characteristics further complicate attempts distinguishing the units in this location. Nevertheless, at the 53 locations where we are confident about the presence of the “Manizales layers”, the thickness of the deposits and the maximum lithic clasts (average of the five largest lithic fragments) of each tephra layer were measured, the last along the major dispersion axis. Physical characteristics of the tephra for each layer were also described (color, structure, sorting, components, pumice size fragments, sphericity, and angularity). The four layers were defined as Lower Manizales (Lm), Middle-lower Manizales (Mlm), Middle-upper Manizales (Mum), and Upper Manizales (Um) and together are defined as the “Manizales tephra sequence” (Fig. 3).

Fig. 3
figure 3

Composite stratigraphic column of the “Manizales tephra sequence” (left), pictures illustrating the tephra layers observed at the outcrop scale (middle), and sketches that illustrate the eruption column heights and ages (right). Lm: Lower Manizales; Mlm: Middle-lower Manizales; Mum: Middle-upper Manizales; Um: Upper Manizales

Full size image

2.2 Radiocarbon ages

AMS radiocarbon dating was carried out on material from the topmost five centimeters of paleosols directly beneath the tephra layers, following standard sample collection (e.g., Avellán et al. 2023). 500 g of paleosol was taken for each 14C analysis that was performed at the Laboratoire de Radiochronologie Centre d´études nordiques, Université Laval, Canada. Ages were calibrated via OxCal program version 4.4 (https://c14.arch.ox.ac.uk/oxcal.html#program) using the IntCal20 calibration curve.

2.3 Physical parameters of the eruptions

2.3.1 Tephra volume and dense rock equivalent

The proximal volume for each eruption was calculated following the empirical method of Pyle (1989; 1995) and Fierstein and Nathenson (1992). The isopach maps helped to establish a line segment and exponential decay slope equation to find each of the extrapolated maximum thickness (To) and constant k (Table 2). Likewise, we used Eq. 18 of Fierstein and Nathenson (1992), combined with the method of Sulpizio (2005), to obtain the k1 and √Aip parameters and then estimate the total volume of the four tephra layers (Table 2).

2.3.2 Column height and eruptive style

The maximum height of the eruptive column was calculated for each tephra layer following the method of Carey and Sparks (1986). We built the isopleth maps with average lithic fragment diameters of 0.8 cm, aided by Fig. 16 in Carey and Sparks (1986). We also used the method proposed by Pyle (1989) to calculate the neutral buoyancy height (HB) below the umbrella region.

2.3.3 Emission rate and eruption duration

The estimated maximum eruption column was calculated after conversion to a maximum emission rate according to the method of Wilson and Walker (1987). For this calculation, we used the eruptive column height values obtained following Carey and Sparks (1986) and Pyle (1989). The minimum duration time of the eruptions was calculated following Wilson (1976).

3 Results

3.1 The Manizales tephra sequence

The Lm, Mlm, Mum, and Um tephra layers crop out west of the axis of the Central Cordillera. In most outcrops, it is very common to find only the three lower layers (Lm, Mlm, and Mum layers) because the Um layer has been eroded. Figure 4 illustrates the stratigraphic relationships of the “Manizales tephra sequence”.

Fig. 4
figure 4

Correlation of the most relevant stratigraphic sections found in the region. Note the basement that corresponds to both the Casabianca Formation and the Quebradagrande Complex

Full size image

Regionally, the tephra sequence is mantling the metasediments of Early Cretaceous Quebradagrande Complex (Villagómez and Spikings 2013) and Pliocene—Pleistocene volcano-sedimentary Casabianca Formation (Borrero and Naranjo 1990). Below, we describe the stratigraphic succession from oldest to youngest (Fig. 4). All tephra layers are separated by paleosols. The tephras can reach thicknesses of up to 60 cm in Manizales city.

3.1.1 Lower Manizales layer (Lm)

In most exposures, the Lower Manizales layer overlies a 20 cm-thick paleosol developed from the Casabianca Formation (p0 in Fig. 3). This tephra layer is preserved with an average thickness of 20 cm at Manizales city (25 km west of Cerro Bravo crater) and mantles a minimum area of 431 km2 within the 10 km-thick isopach. The base of the layer is gray to light gray, while the top is white to brown. In Manizales, Lm layer is presented as a single massive, very well-sorted, clast-supported, and reversely graded layer that varies from fine-grained ash to medium-grained lapilli. The layer mainly consists of white angular to sub-angular pumice, two types of volcanic lithics (gray fresh and red hydrothermally altered fragments), and free crystals. Pumice fragments have spherical to sub-spherical vesicles and contain plagioclase, amphibole, and biotite crystals. The volcanic lithics are angular to sub-angular with aphanitic texture. Free crystals correspond to plagioclase, amphibole, biotite, and pyroxene.

3.1.2 Middle-lower Manizales layer (Mlm)

The Mlm layer is exposed in Manizales city above a 150 cm-thick paleosol developed atop the Lm layer (p1 in Fig. 3). This tephra layer is preserved with an average thickness of 20 cm in the city and mantles a minimum area of 416 km2 within the 10 km-thick isopach. In Manizales, the Mlm tephra layer appears as a white to light yellowish single massive laterally continuous, clast-supported, fine-grained lapilli, moderately to well-sorted deposit. Base and top levels of Mlm tephra show a coarse-to-medium-grained thin ash layer. The Mlm tephra mainly consists of white sub-angular to rounded pumice, two types of volcanic lithic (porphyritic angular to sub-angular gray-fresh and red hydrothermally altered fragments), and free crystals. Pumice fragments present sub-spherical to elongated vesicles with amphibole and plagioclase crystals. Some of these amphibole crystals are broken and present mosaic structure. Free crystals correspond to plagioclase, amphibole, and pyroxene.

3.1.3 Middle-upper Manizales layer (Mum)

The Mum layer mantles a 20 cm-thick paleosol developed atop the Mlm layer (p2 in Fig. 3), with an average thickness of 27 cm in Manizales city. It covers a minimum area of 493 km2 within the 10 km-thick isopach. In Manizales, the Mum tephra layer corresponds to a laterally continuous deposit, which is relatively thick compared to other layers. The tephra unit is moderately to well-sorted and presents a diffuse stratification given by the presence of a thin layer of ash in its middle portion. This layer is formed by fine-to-coarse-grained lapilli that mainly contains ochre sub-angular to sub-rounded pumice fragments, three types of lithics (gray angular fresh fragments and red hydrothermally altered volcanic clasts, and in very low proportion gray metamorphic rocks), and free amphibole, plagioclase, and biotite crystals. Pumice fragments show sub-spherical to spherical vesicles and amphiboles and plagioclase crystals. Likewise, some of these amphibole phenocrysts are broken and present mosaic structure.

3.1.4 Upper manizales layer (Um)

The Um layer in the city overlies a 30 cm-thick paleosol developed on top of the Mum layer, with an irregular contact (p3 in Fig. 3). The Um tephra layer is preserved with an average thickness of 12 cm in Manizales city and mantles a minimum area of 369 km2 within the 10 km-thick isopach. In Manizales, the tephra layer is characterized by a gray-to-light pinkish-gray massive lower horizon and a white upper horizon. This tephra deposit appears as a clast-supported, fine to medium-grained lapilli layer with reverse grading. The Um tephra is well-sorted and contains white angular pumice, gray fresh to red hydrothermally altered volcanic lithics, and free amphibole, plagioclase, and biotite crystals. Pumice clasts present sub-spherical to spherical vesicles with plagioclase and amphibole crystals. The volcanic fragments are mainly concentrated at the base of the deposit.

3.2 Radiocarbon ages

Four dark-brown and clayey humic-rich paleosols located directly beneath each tephra layers were dated (Fig. 5; Table 1). Despite having a relatively irregular stratigraphic contact surface, they are very well preserved against erosion. Laboratory results reveal a modern carbon fraction of 0.4168 ± 0.0011 for Lm, 0.4237 ± 0.0009 for Mlm, 0.5683 ± 0.0009 for Mum, and 0.7195 ± 0.0014 for Um tephra layer, and D14C values of − 583.2 ± 1.1, − 576.3 ± 0.9, − 431.7 ± 0.9, and − 280.5 ± 1.4 ‰, respectively (Table 1). The conventional radiocarbon dating for these paleosols yielded uncalibrated ages of 7030 ± 25 years BP for Lm, 6900 ± 20 years BP for Mlm, 4540 ± 15 years BP for Mum, and 2645 ± 20 years BP for Um tephra layer (Table 1).

Fig. 5
figure 5

Diagrams illustrating the 14C radiometric ages obtained using the OxCal 4.4 program online (https://c14.arch.ox.ac.uk/oxcal.html#program) and the IntCal20 calibration curve. Note the calibrated range values of the ages with 1σ (68.3%) and 2σ (95.4%)

Full size image
Table 1 Radiocarbon ages for p0, p1, p2, and p3 paleosols from the “Manizales tephra sequence”
Full size table

4 Discussion

4.1 Tephra distribution and eruption source

The distribution of the four tephra layers reflects a general westward direction from the Central Cordillera (Fig. 6). The layers were well-recognized in the surroundings of Manizales city and the municipalities of Villamaría and Neira, however, they were difficult to distinguish in the proximal areas where more than a dozen very similar tephra layers appear. Despite this, it was possible to calculate minimum areas for the tephra layers of 431 (Lm), 416 (Mlm), 493 (Mum), and 369 (Um) km2 within the 10 cm-isopach. The mapping indicates that Mlm, Mum, and Um originated from Cerro Bravo volcano, while Lm had an origin further north. Based on the isopach (Fig. 6) and isopleth maps (Fig. 7), we consider Plazuela volcano as the best possible source for this tephra layer. This volcano (Fig. 2B) is located 9.5 km NW of Cerro Bravo volcano and 15 km NE of Manizales city.

Fig. 6
figure 6

Isopach maps. Lower Manizales (Lm) (Top-left). Middle-lower Manizales (Mlm) (Top-right). Middle-upper Manizales (Mum) (Bottom-left). Upper Manizales (Um) (Bottom-right). White dots show the locations where the deposit thicknesses were measured. The slope values obtained for each isopach are observed in the upper left

Full size image
Fig. 7
figure 7

Isopleth maps. Lower Manizales (Lm) (Top-left). Middle-lower Manizales (Mlm) (Top-right). Middle-upper Manizales (Mum) (Bottom-left). Upper Manizales (Um) (Bottom-right). White dots show the locations where the diameter of the pumice fragments were measured

Full size image

4.2 AMS 14C ages

The results of the calibrated radiometric ages obtained are in agreement with the stratigraphic position and fall between ~ 7.9 and ~ 2.7 ka (Table 1). This is consistent with the F14C and D14C values since the lowermost paleosol (p0) presents a lower value of modern coal fraction and a higher value of D14C, compared to the younger overlying paleosols, which gradually change to a higher value of modern coal fraction and lowest D14C value for the youngest paleosol (p3) (Table 1). This indicates that there was no disturbance in the tephra layers overlying the paleosols due to processes of natural erosion or anthropogenic excavation, which yield reliable results. Subsequently, after calibration based on the IntCal20 calibration curve, we have obtained the most probable ages that the paleosols represent (Fig. 8). Thus, the ages within a confidence value of 2σ (95.4%) are the following: for p0 an age of 7879.5 ± 55.5 with 84.2% probability; for p1 an age of 7730 ± 55 with 95.4% probability; for p2 an age of 5294 ± 19 with 37.0% probability and an age of 5145.5 ± 25.5 with 35.3% probability; and for p3 an age of 2759 ± 20 with 95.4% probability. These ages obtained for each underlying paleosol correspond to the closest maximum age of each of the eruptions that deposited the tephra layers studied in this work.

Fig. 8
figure 8

Diagrams to determine some of the eruptive parameters. A Crosswind range (y axis) vs. maximum downwind range (x axis) graph, for clasts with diameters of 0.8 cm (adapted from Carey and Sparks 1986) used to estimate the column heights of the Mlm, Mum, and Um pyroclastic eruptions. B Diagram bt vs. bc/bt (Pyle 1989) used to classify the eruptive style linked to the studied tephra layers based on the diameter of lithic fragments

Full size image

4.3 Tephra volume and dense rock equivalent

Following Pyle (1989; 1995) and Fierstein and Nathenson (1992) as described above, proximal volumes of 0.16 (Lm), 0.25 (Mlm), 0.29 (Mum), and 0.17 (Um) km3 were calculated. Then, after Fierstein and Nathenson (1992) and Sulpizio (2005), total tephra volumes of 0.21 (Lm), 0.32 (Mlm), 0.38 (Mum), and 0.22 (Um) km3 were calculated. Finally, we assumed an average pumice density of 1000 kg/m3 and a dacitic magma density of 2600 kg/m3, yielding a dense rock equivalent (DRE) of 0.04 (Lm), 0.06 (Mlm), 0.075 (Mum), and 0.04 (Um) km3 for the tephra layers.

4.4 Column height and eruptive style

Following Carey and Sparks (1986), column heights of 18 (Mlm), 21 (Mum), and 20 (Um) km, and wind speeds of around 31 m/s for all tephra layers were calculated (Fig. 8). The total column height for the Lm tephra was not possible to determine because the diameter of the lithic fragments did not allow the construction of the 0.8 cm isopleth map as required for the method. The isopleth maps revealed slope k values of 0.124 (Lm), 0.183 (Mlm), 0.23 (Mum), and 0.106 (Um), and half distances of the maximum clast extrapolated (bc) parameter of 3.15 (Lm), 2.14 (Mlm), 1.70 (Mum), and 3.67 (Um). These values yielded neutral buoyancy column heights of 20 (Lm), 16 (Mlm), 14 (Mum), and 23 (Um) km. Subsequently, values of HB/HT ~ 0.7 ratios were obtained (Pyle 1989), leading to total column heights of 28 (Lm), 23 (Mlm), 20 (Mum), and 33 (Um) km. The results of the three upper tephra layers are consistent with values obtained for Carey and Sparks (1986). In addition, the value obtained for the Lm tephra layer is too high, which is not consistent with the size of the lithic fragments.

On the other hand, the isopach and isopleth maps of the deposits helped to calculate the values needed for defining the eruptive style (bt and bc parameters of Pyle 1989) (Table 2). The results of the three lowest tephra layers are coherent with the results required for a Subplinian eruption, except the Um tephra layer which is consistent with a Plinian eruption.

Table 2 Eruptive dynamic values obtained for each of the studied layers
Full size table

4.5 Emission rate and eruption duration

Emission rates, calculated using the eruptive column height values obtained following Carey and Sparks (1986), resulted in 3.38 × 107 (Mlm), 6.27 × 107 (Mum), and 5.16 × 107 (Um) kg/s values. For the Lm layer we used the column height value obtained by Pyle (1989), resulting in an emission rate of 2.16 × 108 kg/s (Lm). Minimum duration time of the eruptions, based on these results, as well as on the dense rock equivalent (DRE) volume obtained from the isopach maps and the stratigraphic survey following Wilson (1976), yielded time values of 8 min for Lm, 1 h with 16 min for Mlm, 51 min for Mum, and 33 min for Um tephra layers.

4.6 Hazard implications

The results obtained in this work, particularly the distribution of the lowest layer (Lm) and the age of the deposits as well as the column height of the eruptions, indicate that Manizales city is not only strongly threatened by events associated with Cerro Bravo eruptions as has previously been considered (e.g. Calvache et al. 1987; Lescinsky 1990; Diaz 1991; Monsalve 1991; Monsalve and Núñez 1992; Gil 2001), but also by other volcanoes nearby. As evidenced in this work, eruptions of other volcanoes such as Plazuela could also pose significant volcanic hazard to the city. We emphasize that other analyses such as wind direction on a regional scale should be considered to have a better idea of the real volcanic hazard for the region. For instance, following the wind data direction found on www.weatherspark.com, from March 28 to October 17 only an eruption that occurred south of the SCVTP could affect the city (e.g., the Nevado del Ruiz volcano), while from December 19 to March 28 eruptions from Romeral, Plazuela, Cerro Bravo, or Nevado del Ruiz volcanoes would strongly affect Manizales. From October 17 to December 19 an eruption from any of these volcanoes would not affect the city. In the event of an eruption affecting Manizales, remobilization of the unconsolidated material would also result in the formation of lahars considering the high level of rain in the region (~ 2000 mm/year). It is also worth mentioning that the water supply for the region is taken from a natural reserve area, which is located between Manizales and the Central Cordillera, where strong affects from tephra are expected. The local airport and the main road from Manizales to Bogotá (the capital city of Colombia) would have both to close. Effects would include damage to crops, collapse of roof tops of buildings, pollution of water sources, mechanical damage of machinery in industrial areas, interruption in air transport, choking of air filter systems of air-conditioning, car filters and temperature regulatory systems of critical infrastructures such as hospitals, food factories, respiratory infections and breathing difficulty, among others.

Regarding the Cerro Bravo volcano specifically, the geological history known so far indicates that it has been active at least for 230 ka (Bustos et al. 2023), with highly explosive eruptions in the last 14 ka (Herd 1982; CHEC 1983; Thouret et al. 1985; Calvache et al. 1987; Lescinsky 1990; Diaz 1991; Monsalve 1991; Monsalve and Núñez 1992; Gil 2001). These eruptions, which have been defined as Plinian and Pelean (Herd 1982), have been classified with a volcanic explosivity index between 4 and 6 (Lescinsky 1990; Monsalve 1991; this work). Due to the location of the populations with respect to the volcano, the event with the greatest potential hazard is a pyroclastic fall. These can reach extensions as large as tens of km, taking into account the height at which the volcano is located (4000 m a.s.l.) and the height of the eruptive columns it generates. In Manizales (25 km away), the most populated city around the volcano, the layers evidence up to 60 cm of thickness and the pumice fragments up to 13 cm in diameter. Both values indicate that the city is located in a high-hazard zone due to pyroclastic falls. Based on this work, it is considered necessary to update the Cerro Bravo volcanic hazard map and create hazard maps for other less-studied volcanoes in the region.

5 Conclusions

The source associated with three out of the four thickest deposits in the Manizales region corresponds to the Cerro Bravo volcano. However, the distribution and characteristics of the lowest tephra layer suggest a different source. This work proposes the Plazuela volcano as the origin of this layer; this volcano is located 9.5 km NW of the Cerro Bravo and 15 km from Manizales city.

The analysis of the deposits indicate that a highly explosive eruption occurred at the Plazuela volcano. These caused the deposition of ash with an average preserved thickness of 20 cm at Manizales city. Close in time (135 years later) another highly explosive eruption occurred at the Cerro Bravo volcano. The eruption provoked the deposition of fine-grained lapilli with an average preserved thickness of 25 cm at Manizales city. After a pause of ca. 2544 years, another explosive eruption took place at Cerro Bravo volcano. The eruption deposited a tephra layer in Manizales city with an average thickness of 20 cm. The last significant explosive eruption recorded in the Manizales region was also from Cerro Bravo volcano around 2700 years ago after a time pause of ca. 2535 years. The plume caused the deposition of tephra with an average thickness of 12 cm.

The distribution of deposits suggests that winds during eruptions predominantly blew westward with speeds of 31 m/s for all tephra layers. In Manizales, the wind typically has this direction from December 19 to March 28. Radiometric ages also show that there is a latent hazard from Plinian-type eruptions for the city of Manizales and other nearby populations.

Finally, it is emphasized that previous lack of correlations of the pyroclastic fall deposits with their source volcanoes in the SCVTP may have resulted in imprecise pyroclastic fall hazard assessments for the region. Because of this, hazard evaluation should consider all the volcanoes from the volcanic chain, and not solely focus on the active Nevado del Ruiz volcano.