Introduction

Lahars are hydro-volcanic processes occurring on volcanic environments. During a lahar event, the transported materials are a mixture of rock, water, and mud that travel at high speed through the ravines of volcanoes (Pierson and Major 2014). These events have a high sediment carrying capacity, and their mobility is related to their proportion of water and sediment (Smith and Fritz 1989). Due to the difficulty to predict lahar events and the lack of territorial planning, they cause every year hundreds of fatalities around the world, as well as a large amount of damages in infrastructure, housing, and agriculture (Thouret et al. 2020), which may jeopardize vulnerable communities living in volcanic zones (UN 2015). The analysis and understanding of lahar dynamics are fundamental to implement disaster risk reduction (DRR) strategies (Chmutina et al. 2019).

Improving the resilience of people living in volcanic regions requires the elaboration of reliable hazard and risk laharic maps (Pierson et al. 2014). To this end, the use of physically based numerical models has become popular since the 1990s, e.g. LaharZ model (Schilling 2014); Titan 2D model (Williams et al. 2008); FLO2D (O’Brien et al. 1993); or RAMMS (Christen et al. 2012). In combination with field-based analyses, different numerical models have allowed to reconstruct recent lahar events and to understand potential impacts by taking in consideration the spatio-temporal variations of physical parameters (Davila et al. 2007; Iverson et al. 1998; Muñoz-Salinas et al. 2009; Williams et al. 2008; Worni et al. 2012). For instance, the LaharZ model was used to reconstruct the lahars events of Ruapehu and Taranaki volcanoes, New Zealand (Stevens et al. 2003); while the Titan 2D was used to develop a lahar hazard map of Merapi volcano, Indonesia (Charbonnier and Gertisser 2012). The hydraulic model FLO2D has been also used to report the magnitude and impact of massive lahar events on Nevado de Huila volcano, Colombia (Worni et al. 2012), as well as laharic activity on Ischia Island, Italy (Nocentini et al. 2015), while the one-phase numerical model RAMMS was used to simulate lahar-like events in Guatemala (Luna 2007), Chile (Bono and Amigo 2015), and Mexico (Franco-Ramos et al. 2020).

Although the use of numerical models has contributed to define flood zones and perform hazard and risk assessment (Pierson et al. 2014), their reliability highly depends on the existing information of the event for model calibration and uncertainty constrains. Thus, gathering evidence-based information from lahar events activity is a real challenge, often constrained to recent events (Thouret et al. 2020). An alternative to gather long-term baseline data on unrecorded lahar events is the use of trees as lahar’s eyewitness (Bollschweiler et al. 2010). Trees growing along lahar channels and slopes of volcanoes are sensitive to such extreme events (Bollschweiler et al. 2010; Pierson et al. 2011), and its analyses allow to (i) precisely define the temporality (annual or seasonal) of past events and (ii) to determine the affected areas along the channels (Franco-Ramos et al. 2013). Thus, botanical-based lahar reconstruction can be used to determine the timing and possible triggers of extreme events, including those of hydrometeorological, seismic, and volcanic nature (Franco-Ramos et al. 2016a, b). Besides, some botanical evidence such as scars on trees represents a paleostage indicator (PSI) of the flow thickness, which can be used to calibrate numerical models (Wilhelm et al. 2019) and indirectly estimate the lahar magnitude (Franco-Ramos et al. 2020). The application of dendrogeomorphological methods for the reconstruction of geomorphic events in volcanoes is in constant evolution (Bollschweiler et al. 2010; Franco-Ramos et al. 2020; Franco-Ramos et al. 2013; Garrote et al. 2018; Salaorni et al. 2017; Stoffel et al. 2011). However, long-term records of the magnitude and temporality of lahars remain scarce worldwide, as is the case for the Pico de Orizaba volcano.

In this study, we aim at providing the longest lahar magnitude reconstruction of the Trans-Mexican Volcanic Belt (TMVB) and as far as we know in the world. Specifically, we focused on a sector of Jamapa Gorge located on the northern slopes of Pico de Orizaba, one of the largest volcanic structures in North America. Analogue to other botanical evidence-based approaches in mountain torrents (Ballesteros Cánovas et al. 2011 Díez-Herrero et al. 2013a, b; Bodoque et al. 2015; Garrote et al. 2018; Tichavský et al. 2020), here we combine botanical evidence from Pinus hartwegii trees growing along the gorge and numerical model resolved in high-resolution digital terrain models (DTM) obtained by unmanned aircrafts vehicles (UAVs) (Fernández de Castro Martínez et al. 2018). Finally, we use the reconstructed lahar record to analyze the potential occurrence of lahars based on data from the closer meteorological stations (Servicio Meteorológico Nacional (SMN) 2008, 2012, 2016) and the hurricane record (National Oceanic and Atmospheric Administration (NOAA) 2019). The results are expected to improve our understanding on the potential magnitude of expected lahars triggered by climatic factors, and therefore to contribute to the implementation of reliable disaster risk reductions strategies in the region.

Study site

The Pico de Orizaba or Citlaltépetl volcano is located on the eastern sector of the TMVB, between the states of Puebla and Veracruz (Fig. 1A). It is a composite cone of andesitic composition with an elevation of 5675 m a.s.l., which places it as the highest volcanic structure in North America (Carrasco-Núñez 2000). The volcano is located on large Mesozoic carbonate sequences of the Sierra Madre Oriental system. Its eruptive history consists of numerous explosive events including Plinian eruptions, sector collapses, extensive pyroclastic flows, and lahars from the late Pleistocene to the early to mid-Holocene, with relatively minor activity in the last few millennia and last observed activity in the late 16th and late 17th centuries (Hoskuldsson and Robin 1993; Siebe et al. 1993; Carrasco-Núñez 2000). Currently, the Pico de Orizaba is considered an active volcano but experiencing a quiet state. Due to its size and geographic position, it represents a major orographic barrier that divides the semi-arid central highlands from the more humid coastal plain of the Gulf of Mexico.

Figure 8A shows the year of the reconstructed lahar, with the occurrence of the hurricane and the recorded precipitation at the closest meteorological station. The lahars seem to be associated with accumulated rainfall of > 230 mm over 3 to 5 days for events in 1999, 2012, and 2016. Yet, these rainfall thresholds seem to be much lower (<130 mm over 3–5 days) in other years, i.e. 1931, 1960, 1975, and 2014, probably reflecting the non-linearity lahar response to environmental and climatic factors. Figure 8 A suggests an increase in lahar formation, going from one or two low to medium intensity phenomena per decade to two in the last decade. Figure 8 B shows the paths of hurricanes formed in the Atlantic Ocean that impacted the coasts of the Gulf of Mexico in the years when there was evidence of lahar formation in Jamapa Gorge. In all cases, the Pico de Orizaba volcano was located in the area of influence of high humidity of 160 km, as per defined by National Oceanic and Atmospheric Administration (NOAA) (2019). This figure also shows that four of the five hurricanes (1931, 1960, 2012, and 2016) come into direct contact with this area of influence. In the case of the 1975 event, the path of the phenomenon is farther away from the volcano; however, it is the only one that reached category 3, according to NOAA intensity scale.

The Jamapa Gorge basin is located between 18° 45′ and 19° 13′ N and 96° 17′ and 97° 16′ W, beginning at the base of the Jamapa glacier at ~4800 m a.s.l. (Fig. 1C). It has an area of about 3912 km2, almost entirely within the state of Veracruz, with an important socioeconomic interest for people (Comisión Nacional del Agua (CONAGUA) 2014). Precipitation in the area ranges from 1200 to 1700 mm per year, with highest values from June to October, according to data from the Huatusco station (30342) (Servicio Meteorológico Nacional (SMN) 2008, 2012, 2016).

Fig. 1
figure 1

A Location of Pico de Orizaba volcano in the Trans-Mexican Volcanic Belt. B and C Satellite image showing Pico de Orizaba and the location of the study area in Jamapa Gorge (Source Google Earth, 2019)

Full size image

Since the late Quaternary, the generation of volcanoclastic flows (avalanches and lahars) is a recurrent phenomenon on the slopes and valleys of the Pico de Orizaba volcano (Fig. 1B). Hoskuldsson and Robin (1993) proposed the existence of four large avalanches responsible for the current volcano morphology. However, Carrasco-Núñez et al. (2006) suggested that only two major collapse events affected the ancestral buildings of the Torrecillas Cone and Espolón de Oro Cone, namely events related to the (i) Jamapa Avalanche Deposit and (ii) Tetelzingo Avalanche-Lahar Deposit (4660 ±100 and 4040 ±80 y.B.P) (Siebe et al. 1993) . In the southeastern sector of the volcano, along the Metlac River, there is evidence of several smaller debris avalanches (Carrasco-Núñez et al. 2006), while in the north-northeastern slope, along the Huitzilapan Gorge, there are deposits of a major lahar that took place in 1920 (“Huitzilapan lahar”), triggered as a compounded effect of an earthquake of 6.5 ML followed by torrential rainfalls (Carrasco-Núñez et al. 2006). Recently, on 5 June 2003, a large debris flow (~peak discharge of 350 m3 s−1) occurred on the southern slope of Pico de Orizaba after a torrential rainfall (Rodríguez et al. 2006).

Methods

Fieldwork

Two fundamental aspects were considered for the delimitation of the study site. The first one is associated with the morphological evidence of the lahars, i.e. lobes, levees, terraces, and the main channel (Fig. 2A). The second one refers to the presence of trees with record of some geomorphological disturbance related to the occurrence of these phenomena in order to obtain samples with dendrogeomorphological potential. Based on the above criteria and the observations made in the field, the selected sector is located in the middle portion of the gorge, from 3630 m a.s.l. to its lower limit at 3530 m a.s.l. It has an area of 0.10 km2 and a length of 650 m (Fig. 2B).

Fig. 2
figure 2

A General view of the studied stretch of Jamapa Gorge showing the main channel filled by recent laharic deposits; a Pinus hartwegii forest covers the adjacent hillslopes and terraces. B Recent laharic deposits that have impacted the trees growing there

Full size image

The fieldwork consisted in (i) obtaining field measurements and topographic data and (ii) sampling botanical evidence of past lahar activity from affected trees. First, we carried out a photogrammetric survey to obtain the topographic information from the studied gorge based on UAV. We used the DJI’s Phantom 4 and Mavic Pro drones, considering a flight’s height of 100 m above the surface and pictures taken with an 80% lateral and transversal overlap (Fig. 3A). In the field, information about geomorphic features such as pyroclastic deposits, lava flows, and exogenous processes (i.e., laharic terraces, erosive landforms, and recent mass movement deposits) was recorded. Then, we recognized all the affected trees growing on each geomorphic feature related to lahar processes along the channel. To minimize the noise signal, only trees with clear evidence of damages by lahars were considered for sampling. These trees showed scars according to the flow direction at the basis of the stem and with sediments and clasts in the surroundings. We avoided elongated scars caused by fallen trees or lightning as well as scars with charcoal remains which may indicate fires processes. Sampling was done using Pressler drills and electric chainsaw to obtain 5 mm diameter cores and cross sections, respectively (Fig. 3B–C). In total, we sampled 78 trees, obtaining 94 increment cores and 40 cross sections of the species Pinus hartwegii. In addition, we sampled 15 trees located in relatively stable and unaffected areas to build the reference chronology to recognize missing rings and determine growth disturbances in samples affected by geomorphological phenomena. In the field, additional information from trees such as the maximum height of the scar and their exact ground position was recorded (Ballesteros et al. 2011; Ballesteros-Cánovas et al. 2016). Figure 3 D shows some of the procedures to acquire field data. These were subsequently used in the calibration of the models generated in the RAMMS software, by means of a Geographic Information System (ArcGIS 10.5: ESRI 2016). To take the measurements described above, the sediment carrying capacity of lahars was considered, as they can transport all sizes of material, from ash to large boulders. Accordingly, it was assumed that the highest point of the impact scar in the sampled trees is equivalent to the flow thickness, thereby defining a PSI for estimating the peak flow of lahars (Ballesteros et al. 2011; Ballesteros-Cánovas et al. 2016).

Fig. 3
figure 3

A Launch of the UAV in the study site. B Increment cores collected using a Pressler drill, next to a lahar-derived scar, on terrace 5. C Tree killed by a lahar of year 1975 on terrace 3; a wedge sample was collected at 49 cm from the ground using a chainsaw. D Example of PSI and data collected for calibration of the models made in the RAMMS program

Full size image

Laboratory analysis

A total of 498 aerial images were processed in the PIX4DMapper software (Pix4D SA, Switzerland) in the following way: (1) loading and aligning the images; (2) generating and filtering the first point cloud; (3) densifying the point cloud; (4) classifying the ground points; (5) constructing the grid and smoothing the model; and (6) exporting the digital elevation model (DEM) and the orthomosaic (Fig. 4A). From the DEM, the orthomosaic, and fieldwork data, it was possible to make the geomorphological map of the gorge at a scale of 1: 1570. We used the classification of the morphogenetic features by Lugo-Hubp (1988). ArcGis 10.5 (ESRI 2016) software was used for their digital mapping.

Fig. 4
figure 4

A Orthomosaic and DEM obtained from UAV-images with 3.3 centimeters of resolution. B Dendrochronological workstation for measurement and analysis of tree-rings. C Example of a sample with evidence of scar in the year 1975. D Diagram and values of one of the flows modeled with RAMMS

Full size image

All dendrogeomorphological samples were prepared according following Stokes and Smiley (1968). Once the samples were dated following standard procedures (Stoffel and Corona 2014), they were measured using a VELMEX (Robinson and Evans 1980) sliding plate connected to a computer and the TSAPWinTM software (Rinn 2003) (Fig. 4B). In this way, the precise width of each ring was obtained to generate the Tucson type file (.rwl), which is used in the COFECHA program (Holmes 1983), giving validity to the co-dating and the reference chronology. In addition, the disturbances identified in each of the samples were dated according to the criteria described by Stoffel and Corona (2014), who group them on the basis of the intensity and kind of reaction presented by the tree (Fig. 4C). When possible, the intra-position of injures within the annual increment growth (i.e., earlywood, latewood, or dormancy period) was used to distinguish the seasonality of the events (Stoffel and Corona 2014).

Numerical modeling with RAMMS

Numerical modeling of laharic flows was performed using the one-phase RAMMS numerical model (WSL, Bühler et al. 2011; Christen et al. 2012; Fischer et al. 2012). This model uses the Voellmy-Salm fluid flow continuum model to simulate the runout of muddy and sediment-laden flows in mountain terrains (Iverson et al. 1998; Williams et al. 2008; Worni et al. 2012). In this model, the flow resistance is described by the Coulomb friction (μ), proportional to the normal stress at the flow bottom, and the viscous resistance turbulent friction (ξ-m/s2), depending of the square of the maximum velocity. The model input data are (i) DEM, (ii) the initial discharge (volume or hydrograph), (iii) Coulomb friction (μ), (iv) turbulent friction (ξ), and (v) flow density (ρ). During the model setup, we used a 1-m resolution DEM derived by photogrammetry, while the assignment of initial friction values was obtained from the laharic events reported in other volcanoes in Mexico, such as the ones analyzed on Popocatepetl (Caballero and Capra 2014; Caballero et al. 2016) and Colima volcanoes (Capra et al. 2010). Accordingly, the values considered were μ= 0.15, which was defined on the basis of field observations and represents the average slope of the deposition zones (Voellmy 1955), and ξ= 400 m/s2, which represents the nature of the flow observed in similar physiographic contexts in Mexico (Caballero and Capra 2014). The input discharge was added as hydrographs; in this sense, it was considered that in tropical environments, the discharge peaks in most of the ravines occur at 60 s, and for the modeling time, 2000 s were assigned, with the intention that the flow would cover the entire study area. Different peak discharges ranging from 25 to 500 m3/s were modelled. The density considered for the flows was 1400 kg/m3 (Table 1). A sensitive analysis was also carried out by varying ±25% friction and density parameters. Finally, the lahar magnitude assignment was based on the minimum average standard deviations between the water depth modelled and the maximum scar height dated for each event. Once these results were obtained, a linear regression was applied to them and the most adjusted input value was identified (Fig. 4D).

Table 1 Parameters used in RAMMS, and values used in the Jamapa Gorge
Full size table

Hydrometeorological triggering of lahars

The reconstructed lahar activity (both occurrence and magnitude) was linked to rainfall data and in particular to the occurrence of extraordinary hydrometeorological phenomena. The linkages with rainfall events were possible due to the seasonal precision on the lahar reconstruction based on dendrogeomorphology. The analysis included data from meteorological stations near the study area (i.e., Coscomatepec; Ixhuatlán and Huatusco Station; Servicio Meteorológico Nacional (SMN) 2008, 2012, 2016). Owing to the lack of meteorological stations at the study site, we used the nearest available meteorological station, which is located 13 km away from the Jamapa gorge. In each database, we calculated the accumulated rainfall in 1, 3, and 5 days, considering the observations made by Capra et al. (2010) and Vàzquez et al. (2014) who associated the development of large volume debris flows with prolonged rainfall events linked to tropical storms or hurricanes. Usually these rainfall events release at least 20 mm h−1 of rain with a maximum intensity of 95 mm h−1 or even higher amounts reaching 400 mm during the 30 h with maximum intensities of 37 mm h−1 (Capra et al. 2018). Once these values were identified, we searched for the possible event that triggered the increase in rainfall, which mostly coincided with the occurrence of hurricanes from the Atlantic Ocean, so their paths were analyzed based on NOAA reports (2019).

Results

Dendrogeomorphic detection of lahar activity

The geomorphological analyses of the Jamapa Gorge reveal intense activity of exogenous processes shaping the pyroclastic deposits and lava flows associated with the andesites-dacites emplaced during the eruptive history of Pico de Orizaba. In the studied sector, six levels of laharic terraces and recent deposits were identified, modelled by fluvial and gravitational erosive processes (Fig. 5). Trees affected by recent lahar activity were growing on the latter geomorphic units, generally in the first and second laharic terraces, while unaffected trees used for the reference chronology were located in higher terrace levels, where the influence of fluvial and gravitational processes on trees was not observed. Samples from 15 Pinus hartwegii trees that did not show evidence of disturbances were used to build the reference chronology. The chronology covers the years 1709 to 2018. COFECHA indicated a significant correlation of 0.513 (P<0.01) with an average sensitivity of 0.291. Overall, 78 trees were sampled showing the average age of the samples trees was 74 years (nobs=78). Disturbed trees allowed the identification of 157 growth disturbances, consisting of scars (51%), growth suppression (35%), and compression wood (32%). These growth anomalies lead to the detection of 8 lahars in the years 1931, 1960, 1968, 1975, 1999, 2012, 2014, and 2016. The lahars of 1975 and 2012 affected a large number of trees, 16 and 26 trees, respectively. Most of the trees with disturbances were in dynamic geomorphological positions such as low terraces and recent laharic deposits. Figure 6 shows the distribution of the trees which allowed the dating of the lahars.

Fig. 5
figure 5

Geomorphological map of Jamapa Gorge, Pico de Orizaba

Full size image
Fig. 6
figure 6

Spatial distribution of tress showing tree-ring evidence of laharic events in the middle sector of Jamapa gorge

Full size image

Estimation of the lahar magnitude

The lahar magnitude estimation was based on 49 trees growing near the main channel and on low terraces. PSI defined by scars heights ranked from 0.032 to 1.35 m, with an average of 0.56 ± 0.32 m. Overall, the deviation between the flow modelled thickness and the height of scars for all reconstructed events ranked between −0.69 and 1.99 m, with an overall average of 0.004 ± 0.49 m (Fig. 7). The model parametrization we used was μ = 0.15, ξ = 400 m/s2, and flow density =1400 kg/m3. The value that was modified for each case was the input discharge of the hydrograph. These and other results obtained are shown in Table 2. The average discharge reconstructed over the last century in the Pico de Orizaba was 174.5 ± 82.5 m3/s (nobs=8). The largest lahar event took place in 2012, with a reconstructed magnitude of 91 m3/s, while the smallest event took place in 2014, with a magnitude of 80 m3/s. Based on the average deviation, the best adjustment between the PSI and modelled water depth was found for the lahar that took place in in 2012 (average deviation = −0.01m, nobs=19, peak discharge = 91 m3/s), while the event of year 1975 presented a less adjustment but not significant (average deviation = −0.04 m, nobs=13, peak discharge =160 m3/s).

Fig. 7
figure 7

Modeled depth of 8 laharic events based on tree-ring evidence. Circles show the standard deviation registered when comparing the height of the debarking and the depth value obtained from the simulation

Full size image
Fig. 8
figure 8

A Relationship between identified lahars and rainfall data, as well as the possible associated hydrometeorological phenomenon. B Path of the hurricanes that were associated with the lahars identified in the study area

Full size image
Table 2 Results of the 8 lahar events simulated with RAMMS and PSI calibration
Full size table

Hydrometeorological triggers of reconstructed lahars

Five out of eight reconstructed lahars were associated with extraordinary meteorological events such as the passage of hurricanes, namely hurricane “San Nicolas” in 1931, “Abby” in 1960, “Caroline” in 1975, “Ernesto” in 2012, and “Earl” in 2016. All these events occurred in the rain season of central Mexico, between July and October, which matches with the seasonality of the injuries observed in the tree-ring records. The other three events could not be linked to any extraordinary hydrometeorological event, because in the seasonal period in which the damage to the growth rings was detected, there is no record of any. However, they were associated with atypical precipitations that did coincide with the period.

Discussion and conclusions

We showed combined botanical and numerical model evidence supporting the longest lahar magnitude records of the TMVB. Our analyses revealed the occurrence of eight lahars since 1930, five of them (1931, 1960, 1975, 2012, and 2016) associated with the occurrence of hurricanes and the other 3 (1968, 1999, and 2014) associated with atypical rainfall (>100 mm accumulated over 3 to 5 days). The reconstructed lahar magnitude varied from 80 to 350 m3/s, with larger events in 2012 (91 m3/s) and 1975 (160 m3/s). This information is highly relevant for DRR purposes, since data about the magnitude of lahars is uncommon in the region.

Dendrogeomorphological analyses using Pinus hartwegii have been conducted in other volcanoes of the TMVB (Bollschweiler et al. 2010; Franco-Ramos et al. 2017; Prado-Lallande 2017). Pinus hartwegii is one of the longest living species in Mexico (Villanueva-Díaz et al. 2010). Yet, the extreme, high elevation environments where this species grows make it very sensitive to disturbances that often cause changes in the growth and the formation of false rings (Villanueva-Díaz et al. 2010). Once these difficulties were identified and observed in the samples collected, it was possible to perform lahar reconstructions with precise dates. Our reconstruction is associated with some events recorded in other volcanoes of the central-east portion of the TMVB. In particular, Franco-Ramos et al. (2017) and Franco-Ramos et al. (2019) report rockfalls at Cofre de Perote and lahars in La Malinche in 1975. In the case of the 2012 event, laharic flows were also dated on Iztaccíhuatl (Prado-Lallande 2017), Popocatépetl (Franco-Ramos et al. 2016a), and La Malinche volcanoes (Franco-Ramos et al. 2019), apparently in connection with Hurricane Ernesto on the Gulf of Mexico coast on August 9, 2012 (National Oceanic and Atmospheric Administration (NOAA) 2019). These matches support the reliability of our reconstruction and suggest that lahar events respond to regional climate triggers.

Thus, the formation of lahars in Pico de Orizaba seems to be related to the occurrence of hurricanes, as previously reported in Popocatépetl (Bollschweiler et al. 2010; Franco-Ramos et al. 2016a), La Malinche (Franco-Ramos et al. 2019; Franco-Ramos et al. 2016b), and Colima (Capra et al. 2010; Franco-Ramos et al. 2013), where the main trigger is a heavy rainfall (20–95 mm h−1) accumulated over a period of no more than 5 days (200–400 mm), this sudden increase in rainfall being usually associated with hurricanes and tropical storms. Our results suggest that in Jamapa Gorge, lahars are linked to rainfall events exceeding 100 mm accumulated over 3 to 5 days. This rainfall threshold was reached in a single day probably associated with the lahar of 2016 (August 5, 2016; daily rainfall: 141 mm). Yet, we detected that in some events (i.e., 1931, 1960 and 2014), the contribution of rainfall (114–127 mm) was not as significant as in other years. Alternative reasons for the generation of debris flows are (1) fire on mountain slopes, which produces soil hydrophobicity and higher runoff leading to debris flows (Cannon and Reneau 2000; Staley et al. 2013); (2) the increase in temperature in high mountain areas that favors glacier melting and the occurrence of sudden water inflows to mountain streams (Chiarle et al. 2007; Morales Martínez et al. 2016); and (3) the occurrence of regional climatic phenomena such as La Niña, which are characterized in central Mexico by atypical and intense rainfall (Franco-Ramos et al. 2016a, b; Vargas et al. 2006). It is possible that these three factors explain the formation of large and widely distributed lahars that are not associated with a hurricane. However, these linkages remain unclear and more evidence is needed to corroborate them.

We applied 2D numerical modeling to reconstruct the magnitude of lahars based on dendrogeomorphological evidences (Baker 2008; Ballesteros Cánovas et al. 2011 a,b; Benito and Thorndycraft 2005; Díez-Herrero et al. 2013a, b). For each event, a map was obtained in which its spatial distribution, the maximum flow rate, and the position of each collected tree were represented, as well as the standard deviation between the height of the debarking and the recreated flow rate value. The maximum depths recreated range from 5 to 11 m. However, it is necessary to be careful in the interpretation of these results, because in some events, the calibration based on PSI of trees shows a wide standard deviation between the debarking and the flow depth, as in the case of the lahars of 1931 (± −0.03) and 1975 (± −0.04). The opposite occurs for 1960 and 2012, when the modeled depth adjusts more precisely to most scars (±−0.01). Although we carefully chose the reach river in term of bedrock, we recognize that the bars located in the upper and lower part of the studied area could be affected by geomorphic changes. These changes may partially explain the variation in the standard deviation, though other factors such the location of trees within the studied reach may affect (Ballesteros-Cánovas et al. 2016; Quesada-Román et al. 2020). We observed that trees located in areas close to the main channel, on the secondary channel, and on low terraces work as a better markers for calibration, while trees located on higher surfaces (stable terraces) or far from the points where the flow probably crosses show a higher deviation. Similarly, another factor that can influence the calibration is the number of trees used. However, this fact was not considered a determining parameter for the interpretation of the results, because there are no elements to assess this issue. The average input discharge was 174.5 m3/s, with a minimum of 80 m3/s and a maximum of 350 m3/s. The peak discharge was set at 60 s and the software modeled 2000 s. The results obtained in terms of speed, volume, and distribution coincide largely with those reported for the Popocatépetl volcano gorges (Caballero and Capra 2014; Haddad et al. 2016; Muñoz-Salinas et al. 2009) and for Colima volcano gorges (Caballero et al. 2016; Capra et al. 2018), as well as with the simulation of the 2012 lahar in the Jamapa Gorge of Pico de Orizaba (Franco-Ramos et al. 2020).

The results and methodology employed in this work have implications in different aspects. The first is associated with the way in which risks from lahars are approached both in Mexico and around the world. Although there are already studies dealing with their behavior and possible scopes are analyzed, there are still gaps such as magnitude, triggers, and reconstructions over long periods of time. All these aspects were considered in this work and that is how the longest reconstructions of laharic flow magnitudes for Mexico and possibly in the world was obtained. Another implication of this work is related to the elaboration of hazard maps for Pico de Orizaba volcano, as the information generated may be useful to update existing maps. Also, it is important to highlight that the methods used in the research can be replicated in tropical areas where active volcanoes are subject to the impact of hurricanes.

We conclude that the use of new technologies such as UAVs, photogrammetry, 2D modelling, and tree-ring evidences in a sector of Jamapa Gorge was a useful and accurate method to date natural phenomena such as lahars. For the study area, 8 lahar events were identified in 1931, 1960, 1968, 1975, 1999, 2012, 2014, and 2016. The lahars of 1975 and 2012 had the greatest distribution and number of affected trees. Their formation was mainly associated with torrential rainfalls resulting from the passage of hurricanes and tropical storms. This research represents a contribution to the study of hydrogeomorphic processes on volcanic terrains in central Mexico by providing discharge reconstructions for a period of almost 100 years, as well as the direct association between the formation of lahars and their triggers. The methods can be applied in other gorges of Pico de Orizaba as well as in other volcanoes with similar conditions.