Classification of slopes on the Comau fjord.
A geomorphological analysis of the Comau Fjord was carried out to identify geohazards that are a product of current landform dynamics and processes. The geological setting of the area includes fractured metamorphic and volcanic rocks forming steep hillslopes in an active tectonic context due to the Liquiñe-Ofqui Fault Zone (LOFZ). Geomorphological and hazard mapping was performed using aerial photographs, GIS geoprocessing and fieldwork in January and May 2016 and February 2017. The susceptibility of landsliding was statistically assessed and validated with the inventory of landslides completed during fieldwork. The triggering of geohazards such as landslides and fluvial floods in the study area is associated with high annual precipitation (>5000 mm annually) with a concentration of rainfall that has increased in the last 50 years. Geohazard mapping demonstrated the potential for rock and earth falls, debris flows and river floods, as well as the potential impact of these geohazards on the area’s intensive aquaculture industry and a main national highway projected for the eastern flank of the fjord. In a geographical scenario of environmental and territorial change, the present and future human occupation of Comau Fjord’s coast constitutes potential hazard and risk conditions for aquaculture infrastructure and highway users.
- North Patagonia
- Comau Fjord
Natural hazards are due to the interaction between environmental and social changes, and the patterns of land use associated with economic development and the processes of urbanization which increase people’s exposure and vulnerability [1, 2, 3, 4, 5, 6]. In such a context, natural risk is also considered to be an unresolved social problem .
Climate change scenarios are a new challenge for the study of natural hazards, risks and adaptation to hazards [8, 9, 10, 11]. For example, during the period between 1900 and 2013, floods were the most frequent natural disaster, affecting more people than any other event of natural origin, , phenomena that intensively affected urban and rural areas alike .
Kappes et al.  suggest that the different types of hazards, scales of analysis, magnitudes of measures and risks constitute a scenario of multiple risks, which are also associated with environmental changes and human impact. Furthermore, human impact also functions as an agent of change in the processes and behavior of morphological systems .
Glacial and periglacial areas are geomorphological systems of high sensibility to climatic agents [16, 17] that are accentuated by the marine influence of the South Pacific Ocean. The climate changes modeled for the sector [18, 19, 20, 21] that add future productive interventions particular to anthropic agents suggest an even more complex future for natural and territorial systems [22, 23, 24].
The most frequent natural hazards in mountainous environments are shallow landslides, associated with a landscape modeled by glacial and tectonic processes, such as those identified in the fjords of Norway [25, 26, 27]. Furthermore, large tsunamis from landslides have also been registered in Norway’s fjords.
The purpose of this study is to establish the geomorphic conditions in the Comau Fjord, the associated natural hazards, and the impacts on vulnerable areas caused by these hazards, including the direct impact on a highway currently under construction and the potential population of the area.
2. Study area
The Gulf of Ancud and the Comau Fjord are part of the Andean landscape, with marked tectonic, volcanic, and glacial activity and climatic-environmental changes from the Pleistocene era to the present. The Comau Fjord is located between the mouth of the rivers/fjords Quintupeu and Vodudahue, covering a distance approximately 24 km long, with a maximum and minimum width of 11 and 2 km, respectively. The flanks of the fjord have a steep gradient (>30°) with heights that reach above 1000 m.a.s.l. (Figure 1).
Furthermore, the Comau Fjord area is scarcely populated (300 inhabitants during the year) with an incipient aquaculture industry oriented to large international markets. The area’s connectivity is poor, with access only by sea (private and with State subsidy). Nevertheless the construction of a highway on the fjord’s eastern flank is projected and currently under design that could stimulate development of the local and regional economy. Because of the incipient aquaculture industry in the fjord, an increase in population and economic activities in the area can be projected, but these activities will be located in an area with threats of natural hydro-meteorological, seismic-tectonic and oceanographic events of an Andean Patagonian fjord with a meso-tidal regime.
The Comau Fjord is located directly on the Liquiñe-Ofqui Fault Zone (LOFZ, Figures 1 and 2), a regional fault system parallel to the Nazca and South American tectonic plate boundary, with activity during the late Cenozoic era [28, 29]. The LOFZ is a seismically active fault system [28, 29] in Chile’s Northern Patagonia, which in 2007 generated a seismic swarm in the Aysén Fjord (45°25′S/72°581′W), with a principal earthquake (Mw 6.2), landslides and tsunamis. The tsunami wave reached 10 m [30, 31, 32]. The Chaitén Volcano (42°50′S/72°32′W) is also associated with the LOFZ. The volcano experienced a significant eruption in 2008 , and its ash was identified in the zone of Comau (Figure 1).
Though detailed geological studies of the area are not available, the geological map of  shows that the predominant lithological units are Jurassic intrusions (diorites, gabbros, monzodiorites), metamorphic rocks of schist with amphibolitic characteristics of the Paleozoic era and Quaternary volcanic rock.
Climatically, the region is subjected to winds from the west and the alternating cold front systems associated with a subpolar low-pressure zone, creating temperate oceanic and subpolar climates . Extreme climate scenarios, from the most optimistic to the most severe (B2 and A1F1) of the IPCC  suggest that at the end of the twenty-first century (1971–2100) there will be a notable advance of Mediterranean climates with mild summers as far south as 46°S , while maintaining a zone of a more oceanic climate type (Cfb). Sarricolea and Figueroa  show precipitation diminishing by more than 200 mm and temperatures increasing up to 4°C of average annual temperature until the end of the twenty-first century. Using all models of the AR5 for precipitation and temperature, the following information was obtained:
The temperature will increase in the RCP 2.6 and RCP 8.5 scenarios in the same way until 2020. Subsequently in the RCP 2.6 scenario (and until the end of the twenty-first century), temperatures will establish 1.5°C above the average temperatures for the twentieth century. While the RCP 8.5 continues increasing until the end of the twenty-first century with temperatures 3.7°C above those registered in the twentieth century.
Precipitation does not show significant differences between the RCP 2.5 and RCP 8.5 scenarios until 2050, but toward the end of the twenty-first century, the RCP 2.6 maintains the amounts of precipitation, while the RCP 8.5 decreases by 200 mm.
The methodological focus was based on the geomorphology and GIS processes carried out to establish the natural hazard conditions using the following scheme:
Geomorphological mapping: This was undertaken by means of aerial photograph interpretation and complemented with the geological map at scale 1:250,000 . Due to difficult access to the study area, the high slope angle and elevation of both sides of the fjord, boat and helicopter access were used to examine landforms and validate the geomorphological mapping. The fieldwork was done in January and May 2016 and February 2017. A geomorphological classification for Andean cordillera environments was used , combined with lithology and morpho-structure . Furthermore, an inventory of the fjord’s hanging lakes and glaciers was completed.
GIS and proxies for modeling topographic indexes: The methodology of Märker et al. [41, 42, 43] was used to elaborate a high resolution Digital Elevation Model (DEM) (Alos/Prism 10 m o SRTM 25x) hydrologically corrected according to the Planchón and Darboux algorithm , reprocessed for analysis of terrain using SAGA GIS. Processes of hydrological erosion were modeled applying the following indexes: the Stream Power Index (SPI), the Transport Capacity Index (TCI) and the Topographic Wetness Index (TWI). The SPI was used to identify susceptibility to erosion and transport of sediments frequently accumulated in turbulent flows. This model describes the effects of the processes of fluvial lineal erosion and stream incisions, such as gullies, ravines and lateral erosion of beds. The Transport Capacity Index (TCI) was used to indicate areas prone to laminar erosion, sediment transport and processes of deposition. The Topographic Wetness Index (TWI) provides information on the accumulation of water and soil saturation around flat terrain units or topographical depressions, suggesting processes of surface runoff with substrate saturation and areas susceptible to flooding. This index was used to estimate inundation zones along fluvial riverbeds and ravines. Furthermore, modeling of topographical data using GIS tools was carried out to complement the geomorphological analysis, especially for those parts of the mountain on which fieldwork could not be done.
Landslide inventory: A photointerpretation of aerial photographs (1982, 1997, 1:20,000 and 1:70,000) and Google Earth was completed and validated by fieldwork. Landslides were identified by morphological evidence, according to the criteria of Náquira , Sepúlveda and Serey  and Sepúlveda et al. [32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47] for fjord environments. Vegetated landslides were also identified that occurred in the past but at unknown dates.
Landslide susceptibility: The analysis of landslide susceptibility was carried out using the bivariate statistical analysis of Van Westen , applied by Molina , using as variables the gradient, orientation and height of the hillslope, curvature and profile, distance to the drainage networks and to the faults, density of drainage and faults/lineations, and the lithological units as evidence of geomorphological, hydrologic and geological characteristics of the fjord. The method requires a landslide inventory. The landslide susceptibility map was developed using weighing factors according to the method of Dahal et al. , dividing the factors in representative classes of the study area and intersecting with the inventory results. Weights were assigned as a function of landslide density found in each pixel factor. The ranges were reclassified for the study zone by Náquira  as high susceptibility, with 35% of the highest weighted values; moderate susceptibility, represented between 35 and 62%; and low susceptibility, between 62 and 100% of the weight .
Gutenberg-Richter Law and seismic activity: This was used to identify seismic interplate continental or cortical events occurring in the area (41°27′- 43°30′S), according to the data from the USGS in the period 1919–2016. The model connects the frequency and the magnitude of earthquakes . This information is used only as an indicator of recurrent superficial seismic activity associated with the LOFZ and as a potential trigger for landslides and tsunamis.
The results of the Geohazard map are subject to uncertainty given the variability of precipitation and temperature in a climate change scenario, which was not modeled in this study.
4. Results and discussion
4.1. Geomorphological mapping
The geomorphological map shows a predominance of hillslopes of metamorphic and plutonic rock  that have been classified as active slopes because of the strong lineal incision, the presence of free-faces with several discontinuities such as joints, faults and fractures creating planes of weakness around rocky slopes in the zone of seasonal nivation, and the presence of talus, covered by dense vegetation of
|Slope||Area km2||Area %|
|Very strong to moderately steep (20.1°–30°)||60.92||24.99|
|Very steep (>45°)||36.4||14.93|
|Total area km2: 243.84|
Numerous individual cirques and systems of coalesced cirques on both flanks of the fjord are the inherited forms of glacial excavation. The flanks’ altitudinal slopes oscillate between ~600 and ~1000 m.a.s.l. (Figure 2) with respect to the base of the fjord. Landforms indicating active hillslope instability are debris cones and localized slopes at the base of the cirques as well as on the hillslopes of the fjord.
Relevant factors of geomorphology are the numerous fractures and structural lineaments related with the LOZF that are identified in the granite, metamorphic and volcanic rock (Figure 2). These factors were incorporated in landslide modeling.
With regards to alluvial processes, the geomorphological map (Figure 2) shows that alluvial fans on the riverbeds of the fjord are almost inexistent given that the gradient and morphology of the fjord are unfavorable for their development, except for three hydrographic basins, with surfaces over ~15km2, that discharge on the western shore of the fjord. The size of these alluvial fans ranges from ~300 to ~750 m (i.e. linear distance from the apex to the distal zone). On the other hand, in the eastern zone of the fjord, macro tidal fan deltas have formed associated with the three principal Andean hydrographic basins of the Cahuelmó, Huinay and Vodudahue rivers. These are large alluvial fans in a macro tidal environment, formed by coalesced lobes of gravel , except for Cahuelmó, which is itself a delta, according to the predominant sandy sedimentology identified in fieldwork.
Eight hanging lakes were identified on the eastern flank of the fjord on steep slopes with dense native forests (Figures 2 and 3). These are systems of three and four interconnected lakes, filled by a fusion of glacial waters from the receding glacier identified in the last 30 years. These lakes are associated with glacier cirque, located between ~1200 and 1800 m.a.s.l., with a S-SE orientation. In the glacial lakes there are only rocky thresholds, without moraines; these glacial lakes supply water to the hanging falls. There is no glacial lake outburst floods (GLOF) observed.
4.2. GIS and proxies for modeling topographic indexes
The results showed that because of the steep gradient of Comau Fjord’s western hillslope (slope > 45°), high levels of lineal erosion were generated, identified in Figure 4A. The areas with the most significant processes of lineal erosion (red) are principally associated with stream incisions developed on the granite and metamorphic rock slopes with structural lineaments to which are associated the ravines that connect the cirque glaciers and the hanging lakes. In addition, these ravines dissect the fjord walls, permitting the transfer of runoff and detritus from the cirques up to the fjord’s base. The fluvial valleys of the Vodudahue, Huinay and Cahuelmó present elevated ranges of lineal and lateral erosion, associated with the development of a drainage network of glacial and periglacial Andean catchment areas. By contrast, areas with low SPI show good agreement with ridges of divides and fluvial plains, where the slope tends to be lower than 5°.
The results of the Transport Capacity Index (TCI), in Figure 4B shows the marked influence of the gradient on the slopes and the adjacent valleys; these are areas that present the surfaces most affected by laminar erosion (red colors). These areas show evidence of landslides and talus in the slopes, as a response to the high susceptibility to the soil erosion of the hillslopes. The red lines on the bottom of the valley and ravines show the action of the water. There are marked incisions in the ravines and streams, many of which come from glacier-lake or fluvial valley systems, with distal deposits, alluvial fans or fan deltas. Just as with SPI, areas with low TCI are associated with fluvial plains and fjord divides.
The Topographic Wetness Index (TWI), as can be seen in Figure 4C, shows areas with the highest potential of water accumulation are the valley bottoms of the principal catchment basins (Cahuelmo, Huinay, Vododahué), where the low fluvial terraces (Figure 2) present flood risk. The TWI results in the study area clearly indicate the presence of hanging lakes and cirque glaciers. The western slope of the fjord presents scarce areas of water accumulation because of the steepness of its slope, but a moderate potential for saturation that can be associated with areas prone to landslides. The results of the index’s application were validated in fieldwork by helicopter flight.
5. Geohazards: landslide and floods
The model of landslide susceptibility shown in Figure 5 demonstrates high susceptibility on the eastern side of the fjord, which can be associated with steep slopes of metamorphic and igneous rock, a factor that Oppikofer et al. , Blikra et al.  indicate as highly favorable for this type of phenomena. The model included the density of structural lineaments (faults) and lakes, which contribute to the propensity to landsliding in the eastern zone.
Figure 5 also shows the location of debris flows, which show spatial agreement with the SPI and the TCI (Figure 4), associated with hydrologic action in micro catchment areas and ravines. On the eastern hillslope, the presence of receding glaciers and their associated lakes constitute potential areas for debris flows. There are also a large number of coalesced cirque glaciers. Consequently, the results of landslide susceptibility, geomorphology and GIS Index coincide.
The rock fall features cover large sections of the slopes as evidence of geological processes that have been masked by the vegetation of the austral forest and demonstrate probable synergic action of the LOZF’s tectonic action (Figure 6A and B). The predominant forms are earth and rock slides, observed as much of plutonic as of metamorphic rock (Figure 6C and D). Many landslides were found to be covered by vegetation, principally ferns, pioneer vegetation that date from an indeterminate time in the past (Figure 6E);
The debris flows identified do not present a pattern associated with a specific type of rock, rather they are associated with ravines and streams in micro catchments (Figure 7) and glacial retreat catchments.
Areas subjected to fluvial inundation are identified in the mid and lower sections of the principal valley and sub-catchment area tributaries (Figure 5). The morphology of fluvial terraces eroded by historic fluvial rises is evidence of this activity, above all in the most distal parts where the Holocene fluvial terraces still present evidence of seasonal fluvial action (Figures 5 and 7). The fan delta morphology (Figure 8) that drains the principal valleys also shows evidence of flooding due to tidal changes. The results of the TWI (Figure 4) permitted analysis of areas difficult to access, and these results were also validated by helicopter flight in January 2016.
6. Seismic activity and landslide-tsunami trigger
Results of the calculation of seismic susceptibility using the Gutenberg-Richter Law, identified 16 intraplate continental seismic events in the 97 years analyzed (Table 2). Cortical earthquakes correspond to 14% of the total events. Though the coefficient of correlation between the regression analysis and the data is low (0.83), and does not allow for quantifying the return period, a recurrent condition of seismic activity in the LOFZ can be established. Furthermore, this activity is a factor in the geohazard of landslips and tsunamis, as evidenced by Sepúlveda et al.  in the Aysén Fjord.
|Type of seismic event||N° de seismic events||Magnitude||Depth||Coefficient of lineal regression|
|Intraplate continental seismic events||16||3||7.2||7.8||36.2||2.859||−0.440|
|Interplate seismic events||85||3.4||7||7.8||47.5||4.674||−0.631|
|Intraplate oceanic seismic events||13||3.9||6||52.2||165.4||3.505||−0.585|
|Volcanic seismic events||28||2.9||5.2||0||10||8.539||−1.655|
In the context of potential activation of seismic activity that could generate a tsunami, a map was created with areas affected by flooding, adding three wave heights, according to the standards of the State of Chile and the known event that occurred in the Aysén Fjord in 2007.
Figure 9 shows the areas that are potentially susceptible and vulnerable to flooding by tsunami. The State of Chile has defined 30 m above sea level as the safety zone ; the wave height of 50 m is the maximum level modeled for the Aysén Fjord . To this level, 7 additional meters must be added from high tide as a worst-case scenario.
The risk of landslides in fjords has been studied in the northern hemisphere because of its impact on human life and economic loss [26, 27, 57], but in Chile the region of the fjords constitutes an isolated territory uncoupled from the urban-regional-national system, unpopulated and lacking infrastructure .
This situation is in the process of changing with the construction of a land route to foment inter-regional integration. The route is projected to benefit the growth of the aquaculture industry, above all for the export of salmon and the development of economic activity in the valleys of the Huinay and Vodudahue, where small settlements already exist (Figure 9). The case of the Vodudahue is more complex because the valley is quite wide, apt for forestry, livestock and farming, and has internal roads to connect to the route under construction (CH 7), that will connect to the port, continental Chile and South Patagonia. Fishing lodges and an ecotourism industry are being constructed for an elite socio-economic international market.
The predominant feature of Comau Fjord’s geomorphological setting is its steeply graded hillslopes, 36% of its territory in thresholds above 30° (61% have a 20°–30° slope and more than 45° of incline), in addition to fractured intrusive and metamorphic rocks, most of them classified as free-faces and talus landforms, above all on its higher eastern flank. The talus are vegetated providing evidence of a timberline 1000 m.a.s.l., above which are bare rocks exposed to glacial rainfall activity, and processes of nivation. The presence of vegetated talus on both sides of the fjord is a constant, as are the numerous ravines because of intense precipitation and glacial melting activity in the fjord.
The landslide inventory performed in fieldwork is geomorphic evidence of the dynamic processes acting on the hillslopes. The landslide susceptibility modeled for the study area, based methodologically on the inventory and statistical analysis, strongly suggests that the study area shows conditions for landsliding. Furthermore, the seismic activity identified for this zone as part of the influence of the LOZF was described based on the magnitude-frequency relationships using the Gutenberg-Richter Law. Although results show a low coefficient of linear regressions (−0.4) for cortical seismic activity near the study area, this can be considered as another latent trigger for landslides and tsunamis. Historical reports, in fact, have documented the triggering of landsliding inducing tsunamis in Aysen 2007.
GIS modeling of topographic indexes is an important methodological tool for geomorphological terrain analysis in areas with difficult access to the Andean fjords and catchment areas of the Chilean Patagonia. This tool has been useful to show areas prone to be flooded and zones with landslide susceptibility.
The zone’s climatic conditions of very high rainfall (5000 mm/year), together with the presence of low alluvial terraces, favor the dynamic processes that generate fluvial flooding. Identification of areas of fluvial flooding is associated with high annual rainfall and a winter peak level, but also with the increase in spring and summer temperatures that contribute to the receding nature of the Andean glaciers, and the increasing river flow. The trend of climatic change in the zone is set to continue throughout this century. Identified areas of fluvial flooding correspond to the Holocene terraces, currently not in use. The projected highway in the fluvial valleys will be affected by this threat.
The construction of Route CH 7 will be a strong incentive for economic activity in the fjord area, principally in aquaculture and tourism. As shown in the geomorphological and geohazards maps (Figures 2, 5, and 9), most of the planned highway that will be built at the eastern base of the fjord can be disrupted by rock falls, rock and earth slides and debris flows. Indeed, at least 24 critical zones of exposure can be identified on the hazard maps. Moreover, areas that currently have human settlements are located on geomorphological units such us low alluvial terraces and deltas, prone to flooding. These economic activities will, thus, be exposed and vulnerable to future risks given the identified geohazards. For this reason, direct or indirect mitigation controls such as land use planning should be considered to reduce the risk of disaster at the study site.
We would like to thank Fondecyt for financing the project “Recognizing the hotspot in the periglacial environment of the fjords and interior sea through the integrated evaluation of geohazard drivers, risks and impact on territory resources in the Gulf of Ancud: a methodological contribution” (1151087). We would like to acknowledge Huinay Scientific Field Station for their support with field work and information.