use of remote images to analyze the changes in the ...eprints.lincoln.ac.uk/id/eprint/32711/1/use of...

44
Use of remote imagery to analyze changes in morphology and longitudinal large wood distribution in the Blanco River after the 2008 Chaitén volcanic eruption, southern Chile Héctor Ulloa 1 , Andrés Iroumé 2 , Luca Mao 3 , Andrea Andreoli 4 , Silvia Diez 2 , Luis E. Lara 5 1 Universidad Austral de Chile, Graduate School, Faculty of Forest Sciences and Natural Resources, Valdivia, Chile 2 Universidad Austral de Chile, Faculty of Forest Sciences and Natural Resources, Valdivia, Chile 3 Pontificia Universidad Católica de Chile, Department of Ecosystems and Environment, Santiago, Chile 4 Universidad de Concepción, Department of Forestry and Environmental Management, Concepción, Chile 5 Servicio Nacional de Geología y Minería, Volcano Hazards Program, Santiago, Chile Abstract The 2008 Chaitén Volcano eruption generated significant changes in the channel morphology and large wood (LW) abundance along the fluvial corridor of the Blanco River, southern Chile. Comparisons of remote sensing images from the pre-eruption (year 2005) and post-eruption (years 2009 and 2012) conditions showed that in a 10.2 km-long study segment the Blanco River widened 3.5 times from 2005 to 2009, and that the maximum enlargement was 9 times the original width. Changes in channel width were lower between years 2012 and 2009. The sinuosity and braiding indexes also changed between 2005 and 2009. After the eruption the channel sinuosity was higher and specific river reaches developed a braided pattern, but by 2012 the channel was recovering pre-eruption characteristics. Huge quantities of LW were introduced to the study segment; individual LW per 100 m of channel length were 0.2 and 7.5 in 2005 and 2009, respectively, and more than 3 log jams/100 m were observed in year 2009. Between 2009 and 2012 the quantity of LW was very similar. Statistically significant 1

Upload: others

Post on 30-Jan-2021

0 views

Category:

Documents


0 download

TRANSCRIPT

Use of remote images to analyze the changes in the morphology and longitudinal large wood distribution in the Blanco River following the 2010 eruption of the Chaitén volcano, southern Chile

Use of remote imagery to analyze changes in morphology and longitudinal large wood distribution in the Blanco River after the 2008 Chaitén volcanic eruption, southern Chile

Héctor Ulloa1, Andrés Iroumé2, Luca Mao3, Andrea Andreoli4, Silvia Diez2, Luis E. Lara5

1 Universidad Austral de Chile, Graduate School, Faculty of Forest Sciences and Natural Resources, Valdivia, Chile

2 Universidad Austral de Chile, Faculty of Forest Sciences and Natural Resources, Valdivia, Chile

3Pontificia Universidad Católica de Chile, Department of Ecosystems and Environment, Santiago, Chile

4Universidad de Concepción, Department of Forestry and Environmental Management, Concepción, Chile

5 Servicio Nacional de Geología y Minería, Volcano Hazards Program, Santiago, Chile

Abstract

The 2008 Chaitén Volcano eruption generated significant changes in the channel morphology and large wood (LW) abundance along the fluvial corridor of the Blanco River, southern Chile. Comparisons of remote sensing images from the pre-eruption (year 2005) and post-eruption (years 2009 and 2012) conditions showed that in a 10.2 km-long study segment the Blanco River widened 3.5 times from 2005 to 2009, and that the maximum enlargement was 9 times the original width. Changes in channel width were lower between years 2012 and 2009. The sinuosity and braiding indexes also changed between 2005 and 2009. After the eruption the channel sinuosity was higher and specific river reaches developed a braided pattern, but by 2012 the channel was recovering pre-eruption characteristics. Huge quantities of LW were introduced to the study segment; individual LW per 100 m of channel length were 0.2 and 7.5 in 2005 and 2009, respectively, and more than 3 log jams/100 m were observed in year 2009. Between 2009 and 2012 the quantity of LW was very similar. Statistically significant relationships were found between number of log jams and channel sinuosity and between number of pieces of large wood with sinuosity and channel width. Wood was highly dynamic between 2009 and 2012: 78% of individual pieces and 48% of log jams identified in the 2009 image had moved by 2012. Finally, the supervised classification of imagery associated with ArcMap® tools was tested to identify large wood.

Keywords: Chaiten Volcanic eruption; channel morphologic changes; large wood distribution, supervised classification, Blanco River.

1 Introduction

The Chaitén volcano in Chile generated one the major global eruptions since the 1990 (Major and Lara 2013) severely affecting the riparian forests and drainage network of the Blanco river basin (Swanson et al. 2013; Pierson et al. 2013).

Although natural disturbances such as debris flows, floods and forest fires can affect the river basin dynamics, explosive volcanic eruptions have definitely the potential of exerting severe impacts such as filling river valleys, redefining watershed divides, changing river network patterns, channel size, shape, pattern and channel structure because of the huge deposition of tephra and volcanic debris (Manville et al. 2005; Pierson and Major 2014; Zheng et al. 2014; Wisner 2004). These landscape disturbances can strongly alter river system hydrology and sediment transport dynamics with serious consequences on human settlements and transportation infrastructures (e.g., Oppenheimer 2003). Particularly, the major storage components of the water balance may be affected, and the features, timing, magnitude, and duration of runoff strongly altered (Major 2004). These landscape disturbances can trigger accelerated erosion, change the supply of sediment to rivers, and alter the nature of river discharges competent to transport sediments. Commonly, these disturbances can alter runoff hydrology and channel hydraulics. As a consequence, flood events can transport higher volumes of sediments because of the easily erodible river banks and the virtual unlimited sediment supply conditions (Major et al. 2000; Moody and Martin 2001; Hayes et al. 2002). Accelerated erosion and extraordinary sediment transport have been particularly evident after explosive volcanic eruptions that substantially altered landscape hydrology and geomorphology (Major et al. 2000; Inbar et al. 2001; Hayes et al. 2002; Manville 2002). Generally, sediment transport in near-equilibrium fluvial systems occurs during moderate to large floods, but substantial post-eruption transport has been observed under relatively low discharges along highly disturbed channels (Hayes et al. 2002). The amount of sediments supplied from tephra covering hillslopes diminishes rapidly after rill networks expose permeable substrates and slope stabilize (Major et al. 2000). As a consequence, abnormal sediment discharges from basins having mainly disturbed hillslopes decline rapidly and return to pre-eruption levels within tens of months, whereas abnormal sediment discharges from basins having severely disturbed channels can persist for decades. Gran and Montgomery (2005) found that even after the occurrence of major lahars, sediment yields in some basins remain orders of magnitude above pre-eruption levels. After an eruption, river channels also suffer complex cycles of incision, aggradation and widening (Meyer and Martinson 1989), as well as changes in the morphologic pattern (Pierson et al. 2011). Channels can be aggraded by tens of meters and widen by hundreds of meters within months or even during single storm cycles (Pierson and Major 2014). Hydrologic, sedimentologic, and geomorphic responses to major explosive eruptions are usually dramatic, widespread and persistent, and present enormous challenges to those entrusted with managing disturbance responses.

Effects on the fluvial geomorphology and geologic and ecologic processes have been reported after the eruptions of the Mount Saint Helens, USA (Dale, et al. 2005; Major 2003; Major et al. 2009), Mount Pinatubo, Philippines (Hayes et al. 2002; Gran and Montgomery 2005), Mount Etna, Italy (Branca 2003), Mount El Chichón, Mexico (Inbar et al. 2001), Mount Popocatépetl, Mexico (Tanarro et al. 2010), Mount Taupo, New Zealand (Manville and Wilson 2004), Mount Hills, Montserrat Island (Cole et al. 2002) and Mount Misti, Peru (Harpel et al. 2011). With the exception of the work carried out by Lisle (1995) reporting the effects of large wood material (LW) in channels affected by the eruption of Mount Saint Helens, the processes of LW recruitment, abundance, spatial distribution and mobility in the fluvial corridor are very little explored when analyzing the geomorphologic effects after the eruptions. However, better understanding of the variation and distribution of LW is fundamental to comprehend the river channel behaviors (e.g., Piegay and Marston 1998). It has been demonstrated that, generally larger channels and channels draining larger basins contain lower volumes of wood per unit area than narrow channels or low order catchments (Marcus et al. 2002, Abbe and Montgomery 2003, Wohl and Jaeger 2009, Wohl and Cadol 2011, Rigon et al. 2012). However, Rigon et al. (2012) mentioned that volumes of wood material found in mountain streams are highly dependent on landslides processes. Beside, Fox and Bolton (2007) showed that in some cases channels draining larger areas and wider channel feature greater volumes and quantities of wood per unit channel length. In Chile, research on in-stream LW has grown since 2005 (Andreoli et al. 2007a, 2007b, 2008; Comiti et al. 2008a; Mao et al. 2008a, 2008b; Iroumé et al. 2011, 2014; Ulloa et al. 2011), but this set of studies were mainly concentrated in low order channels (basins < 18 km2) not affected by volcanic eruptions.

Field surveys covering large areas of a river channel to characterize morphological changes or evaluate the dynamics of in-stream large wood following a natural disaster are in general very costly and time-consuming, generate data with low spatial resolution, and are not easy to repeat afterwards (Buckley et al. 2000; Marcus et al. 2002; Charlton et al. 2003; Leckie et al. 2005; Comiti et al. 2008b). However, high resolution satellite images (Buckley et al. 2000; Bertoldi et al. 2011) and LiDAR surveys (Cavalli et al. 2008) are becoming more available and affordable, allowing obtaining high resolution digital terrain elevation models. For instance, remote sensing images are used in fluvial geomorphological studies for the assessment of surface grain sediments, large wood and stream habitats (Leckie et al.2005), the spatial analysis of large wood (Marcus et al. 2002), the classification of riparian habitats (Smikrud et al. 2008; Woll et al. 2011), the evaluation of debris flows after Bull et al. (2010) and the recognition of types of river morphology (Cavalli et al.2008; Snyder 2009).

The effects of natural disturbances, especially those related to volcanic eruptions were little studied in the densely vegetated and steep forested watersheds of Southern Chile. Also, the vulnerability of these watersheds to the spectrum of disturbance processes, and the potential hydrologic and geomorphic responses to these phenomena were not well understood. However the May 2008 explosive eruption of the Chaitén volcano concentrated the interest of different international groups of researchers which have studied the chronology of the eruptive processes (Lara 2009; Alfano et al. 2011; Major and Lara 2013; Pallister et al. 2013) and their impacts in the forested slopes and drainage networks (Pallister et al. 2010; Major et al. 2013; Pierson et al. 2013; Swanson et al. 2013). Nevertheless, these studies have considered neither the morphologic evolution of the Blanco river channel (draining the affected area) nor the behaviour of the in-stream large wood material following the eruption.

This study investigates the changes in the morphology and the distribution of in-channel large wood generated by the 2008 Chaiten volcanic eruption in a channel segment of the Blanco (or Chaitén) River. A sequence of remote sensing images was used, one taken just before the eruption and two from years 2009 and 2012. The research focuses on identifying and studying the morphological evolution in terms of fluvial type and channel geometry, and the volume of large wood and its distribution along a 10.2 km-long study channel segment divided into 13 reaches. Furthermore, relationships between morphologic characteristics and their evolution with variables associated with large wood spatial distribution were generated at the reach scale. Finally, in-channel LW abundance was analyzed through supervised classification of imagery and compared with traditional identification.

We tested the hypothesis that (1) pyroclastic flows and lahars generated by eruptions would widen the channel, increasing sinuosity and braiding especially on the lower gradient reaches, (2) between 2009 and 2012 ordinary streamflow processes would entrain wood and sediment from upper to the lower portions of the study river system, (3) that between 2009 and 2012 the downstream reaches would be expected to gain wood from upstream reaches but also from local active channel migration on areas of buried dead trees, and (4) that ordinary streamflow processes would be expected to rearrange the individual pieces of large wood, increasing in the number of wood jams.

Considering the magnitude of the eruption, the impacts on the forested slopes and drainage network and the almost unlimited sediment and wood supply conditions we expect that the Blanco channel will continue to experience important morphologic changes with a period of adjustment extending beyond the study period.

2. Study area

The study area corresponds to a 10.2 km-long channel segment of the drainage network of the Blanco (or Chaitén) River. This is located about 254 km south of Puerto Montt city in the Chilean Lake District (Fig. 1). The Blanco River is part of one of the two main drainage basins connected with Chaitén Volcano (72°39'7,4"W 42°50'1,2"S) through its tributary called Caldera Creek by Major and Lara (2013), Major et al. (2013) and Pierson et al. (2013), also better known as the ‘Ash Creek’ by locals. The Blanco River is located on the southern slope of Chaitén Volcano and runs toward the Pacific Ocean through the town of Chaiten, which is 10 km away from the volcano. If closed at the downstream end of the study segment the basin has a drainage area of approximately 70 km2, an average slope of 50%, and the altitudes range between 7 and 1,545 m.a.s.l. with an average of 616 m.a.s.l. On the other hand, the subbasin of the Caldera Creek tributary has an area of approximately 10 km2, average slope of 55 %, and average, minimum and maximum altitudes of 643, 117 and 1,113 m.a.s.l., respectively. The Blanco River is a fourth-order stream and the regime is dominated by rainfall with winter peak flows. Bedrock geology in the basin is mostly characterized by a Pleistocene volcanic cover overlying a basement formed by both Miocene granitoids and Paleozoic schists and gneisses (Piña-Gauthier et al. 2013). The Blanco River valley is glacially incised in a tectonic lineament, which is thought to be related to the Liquiñe-Ofqui Fault Zone forming a pop-up structure with the main strand (Piña-Gauthier et al. 2013). Soils in the area are mostly originated from volcanic sediments deposited on the bedrock basement. Alluvial and fluvial deposits can be found in the lowland areas where they are interbedded with pyroclastic deposits. A geomorphological feature of these soils is their instability, which favor creeping and frequent landslides (Peralta 1980). As stated by the Chilean Meteorological Office (www.meteochile.cl), the climate in the area is a "west coast cool temperate climate with a winter rainfall" exceeding 3,000 mm/year. The forests cover is of the Evergreen Forest Type (Donoso 1981), characterized by high species diversity and the presence of evergreen tree species such as Nothofagus dombeyi, Nothofagus nitida, Nothofagus betuloides, Luma apiculata, Drimys winteri, Eucryphia cordifolia and Aextoxicon punctatum. According to CONAF (1997) data, 43% of the Blanco River catchment is covered by old growth forests, 40% of shrub forest and 16% of snow and glaciers in the upper areas.

The 10.2 km-long study segment of the channel is divided in two main sections. The 4 km-long upper section (longitudinal slope 12.5%) corresponds to the Caldera Creek flowing from the south of the caldera wall towards the main channel of the Blanco River. The lower section of the segment (6.2 km) is part of the main Blanco River channel characterized by a lower longitudinal slope (1.3%) and an open valley about 350 m wide on either side of the channel. The downstream end of the segment is just upstream of the town of Chaiten to avoid the potential effects of the hydraulic works that have been carried out to protect the town. The two study segment are fundamentally different in terms of slope, being steeper the first one near the caldera (Caldera Creek), and flatter the latter (Blanco River).

The Blanco River segment conveyed pyroclastic density currents and lahars generated by the eruption of the Chaitén Volcano initiated on May 2, 2008 (Carn et al. 2009; Lara 2009). The 2008-2009 eruptive cycle of the Chaitén Volcano is described by Pallister et al. (2013) and Major and Lara (2013). During the eruptive phase of the volcano two main sediment supply events occurred; the initial explosive phase and the subsequent partial collapse of the growing dome, both severely affecting the fluvial corridor of the Blanco River. The first event occurred between June and November (2008) and the second one on February 19th 2009 (Major et al. 2013). The rainfall, registered few days after the onset of the explosive phase, remobilized the pyroclastic cover down to the valley and triggered important sediment yields as lahars or hyperconcentrated flows that became part of the fluvial deposits found in the Blanco River channel (Major et al. 2013; Pierson et al. 2013). Pyroclastic flows, landslides and lahars caused by the partial collapse of the dome and the eruptive column as well as seismic activity (magnitudes between 3 and 5 Mw) generated several changes along the channel (Carn et al. 2009; Lara 2009) and in the riparian forests (Swanson et al. 2010). Major et al. (2013) document from field observations that these strongly channelized flows along the Caldera Creek undermined the stream bed leaving fallen trees. Up to 8 m high deposits on the downstream end of the Caldera Creek are the result of a variety of processes described in detail by Major et al. (2013) and Pierson et al. (2013). The Blanco River valley was highly exposed to the pyroclastic and fluvial flows, which affected directly few km2 of evergreen forests that suffered toppled and transported trees (Swanson et al. 2013). However, trees were not burnt or slightly burnt because pyroclastic flows had moderate temperatures (Major et al. 2013). Swanson et al. (2013) reported that extreme runoff from the upper Blanco River aggraded the river itself, entrained and transported large wood, which contributed to lateral channel change, and deposited up to several meters of reworked tephra, alluvium, and wood on floodplains along the river. Rainfall after the eruption triggered lahar-floods which moved huge volumes of sediments and generated by May 11, 2008, a 7 m aggradation of Blanco River channel near the city of Chaitén (Pierson et al. 2013). Sediments and large wood (LW) blocked the main channel and caused the channel to avulse into the city located some 10 km from the volcano, which had to be evacuated (Carnet al. 2009; Lara 2009; Pallister et al. 2010; Pierson et al. 2013).

FIGURE 1

3. Materials and methods

Characterization of the study segment: The 10.2 km-long segment was divided into 13 individual reaches and numbered from downstream to upstream. Homogeneous reaches were identified according to changes of longitudinal slope as derived from a detailed 2009 digital terrain model, and changes of planimetric features as sinuosity and braiding index (Rosgen 1994).

The 1 m-resolution digital terrain model was obtained from a Harrier 56 airborne LiDAR which included images taken from a digital camera. This survey was contracted by the Los Lagos Regional Government and the processed data was trespassed to the authors through an agreement for research purposes. The Chilean WGS84-SIRGAS official reference system was used and the survey was supported with five vertices distributed in the measured area and positioned with a geodesic satellite geo-receptors. Final precision was estimated around ± 7 mm (Digimapas-Chile, 2009).

Reach lengths range from 400 to 1400 m, which correspond to lengths of approximately 10 to 20 times the width of the original channel on its 2005 conditions. The longitudinal profile of the study segment along with the subdivision of reaches and their slopes is shown in Fig. 2. Reaches 1 to 7 correspond to the downstream part of the segment, in the main channel of the Blanco River, characterized mainly by a low slope and a wide valley bottom. Reaches 8 to 13 correspond to the tributary Caldera Creek, which is a steep area of the basin where pyroclastic flows were confined.

FIGURE 2

Imagery: A multitemporal sequence of three images was used in this study (Fig. 3). The first image, which was obtained from Google Earth®, refers to the pre-volcanic eruption period (Digital Globe satellite image dated 2005, natural colour and 0.6 m spatial resolution) which allowed the identification of the original channel and the extent of riparian vegetation before the volcanic eruption. On this image, the old growth native forests that characterized the pre-eruption condition could be straightforwardly identified. It was also possible to clearly discriminate the original watercourse and the width of the active channel. Additional three band image (RGB, 0.5 m resolution) was derived from an airborne LiDAR survey was taken between 15 October and 7 December 2009, that is after the eruption and the last pyroclastic flow dated February 2009. Finally, a Bundle satellite image (Pan + Mul 4-band with a spatial resolution of 0.6/2.4 m) taken on January 2012 was considered in this study. The analyses were performed using ArcGIS® tools. Georeferencing, correction and orthorectification processes were carried out by the imagery suppliers.

FIGURE 3

Geomorphologic assessment: The width of the active channel and of the active floodplains adjacent to the active channel were manually digitized in the three images. In the 2005 image, the width of the active portion of the river corresponded to the distance between the lines of the riparian forest in both sides of the channel. Instead, in the images of 2009 and 2012, the active portion of the river was considered to be the width of active channel plus the active floodplain bounded by terraces too high to be inundated when the aerial images were collected (hereafter active channel). Relatively little secondary channels flowing into the former floodplain and areas with vegetation overthrown by pyroclastic flows were not considered as part of the active channel. The borders of the active channel were then used to obtain a description of the general channel geomorphology, braiding pattern (i.e. single or multiple channels), and sinuosity (Diaz and Ollero 2005).

Cross sections were drown approximately every 130 m along the main channels, and were used to obtain the average channel width for each channel reach. A total of 77 cross sections were generated, with a minimum of four sections in shorter reaches.

The braiding index (Rinaldi et al. 2010) was measured by counting the number of active channels separated by bars in every cross section (Table 1). The sinuosity index was calculated in every reach with the total sinuosity method (Horacio 2011), considering the ratio between the longitudinal distance measured along the thalweg and the length of the straight-line from the upstream to the downstream end of each reach (Table 1). All these indices were compared to evaluate the planimetric evolution of the channel during the study time-periods.

Table 1. Classification of channel types considering the sinuosity and braiding indices, adapted from Rinaldi et al. (2010).

Typology

Braiding Index

Sinuosity Index

Straight

1 – 1,5

(usually near 1)

1≤ Is<1,1

Wandering

1 – 1,5

(usually near 1)

1,1≤ Is<1,5

Meandering

1 – 1,5

(usually near 1)

≥1,5

Transitional

1 – 1,5

Any

Braided

≥1,5

Any (usually low)

Large wood: All pieces of large wood within the active channel along the 10.2 km-long river segment were identified in every image by digitizing every individual large wood pieces as polyline and the log jams (two or more LW pieces in contact) as polygons. Because individual pieces of large wood shorter than 1 to 3 m-long are not clearly visible in the photos due to the limited imagery resolution, instead of the general minimum length of 1 m (e.g. Iroumé et al. 2014), in this work LW was defined as every piece of wood longer than 3 m (as in other works, e.g. Moulin et al. 2011), The diameter of LW pieces was not considered. This probably led to an underestimation of the total channel LW material. However, it is assumed that the larger and longer individual pieces of large wood and the bigger wood jams are those exerting the most relevant effect on channel morphology and water flow. Subsequently, a comparison of the number of LW individual pieces and log jams observed in every image and an analysis of LW longitudinal distribution and morphologic changes was carried out. For each year and each reach, the spatial density of wood material was calculated, in number of individual pieces of large wood and number of log jams per kilometer of channel length and per active channel area. Comparisons of individual wood length and area of wood jams were also done between 2009 and 2012. In addition, individual wood length and area of wood jams were analyzed comparing LW that were displaced to those ones that remained at the same place.

Furthermore, in reaches characterized by relatively low and high amount of LW (2 and 3, and 5 and 6, respectively), an automatic LW pieces and log jams, a LW identification based on a supervised classification of imagery was performed using an ArcMap® tools based on multispectral imagery obtained from LiDAR data. The outcomes of this classification were compared with results obtained from the manual identification of LW on each image. The supervised classification is a method that assigns different values to an area according to its spectral behavior, with the possibility to choose the number of categories or grouping classes. Additionally, a post-process was carried out to reclassify or eliminate some obvious mistaken polygons, in order to minimize classification errors. Fig. 4 shows a schematic sequence of this process. This method has been previously implemented by Marcus et al. (2002), Leckie et al. (2005) and Smikrud et al. (2008) among others.

FIGURE 4

Relationships between morphological characteristics and evolution with LW redistribution variables: Channel width (m), sinuosity and braiding indices, slope (%, only for 2009) and cumulative channel area (ha) were obtained at the reach scale from the different images. The catchment area at the upstream end of each reach was also calculated. Relationships were analysed between reach channel width and sinuosity, and also between change rates of the maximum reach channel width and sinuosity (width 2009/width 2005 and width 2012/width 2009 with sinuosity of 2009). The wood material was characterized at the reach scale in terms of number of LW pieces and log jams (both in N ha-1 and N km-1), planimetric area of LW jams per channel area (ha ha-1) and maximum size of logjams (ha). Further analysis considered correlations between LW features such as the number of log jams and pieces of large wood per km or ha of active channel, the surface of log jams per ha of active channel and the maximum size of jams as dependent variables, and sinuosity, channel width, accumulated or incremental surface of active channel and catchment area at reach scale as independent variables. Stepwise regressions were used to examine relationships between morphological characteristics and LW redistribution variables. Variables were tested for normality or otherwise log10 transformed. Relations were considered statistically significant at a 90% confidence level.

4 Results

Morphologic changes: The Chaitén volcanic eruption resulted in significant changes of the channel morphology. From 2005 to 2009, channels widened by a factor of 2 to almost 7 times their initial width, whereas from 2009 to 2012 there was much less widening.  In contrast, sinuosity did not respond consistently to widening: in some cases sinuosity increased, and in other cases, it decreased in response to channel widening.  In some cases (reaches 2,3, 4) there was little to no response of sinuosity  to channel widening, and a braided channel was created instead (Table 2). The minimum and maximum reach channel widths were 15 and 105, 31 and 397 and 35 and 397 m during 2005, 2009 and 2012, respectively. Total active channel area for the entire study segment was 35, 107 and 115 ha, and on average 3.2, 8.3 and 8.9 ha km-1 for 2005, 2009 and 2012, respectively (Fig. 5).

If the cross sections are considered, the higher variation in channel width between years 2009 and 2012 was 119 m (1.6 times the pre-eruption width), and only in four cross sections the variation was higher than 0.5 times. However, at the reach level three reaches widened 20% between 2009 and 2012, while the rest varied less than 10%. All differences were statistically non-significant.

FIGURE 5

Before the eruption, the entire study segment featured a single thread channel, being 8 out of 13 reaches wandering and 5 reaches straight. In 2009, which is the closest photo available after the volcanic eruption, 3 reaches featured a braided (i.e. multiple channels) pattern, 1 meandering, 5 wandering, and 4 straight morphologies. By 2012, the channel geometry was more similar to the pre-eruption condition with 9 wandering and 4 straight channel reaches, and the river returned to a single channel pattern (Table 2). Steep reaches increased in width by a similar factor (2 to 3 times) as less steep reaches (1 to 4), but these very steep reaches did not produce braided channels. No relationship was found between changes of channel width and reach slope.

A weak not statistically significant positive relation was found between channel width and sinuosity (Fig. 6), but if the outliers are removed (shown as circles in Fig. 6), the relationship is statistically significant for 2009 and 2012 (R2 = 0.46; P = 0.02 and R2 = 0.78; P < 0.001, respectively).

Table 2. Morphological characteristics of the study reaches.

Reach

Length (km)

Slope (%)

Mean active channel width (m)

Sinuosity index

Braiding index

2005

2009

2012

2005

2009

2012

2005

2009

2012

1

1.0

0.76

33

57

62

1.04

1.04

1.05

1.0

1.0

1.0

2

1.2

0.58

44

97

120

1.22

1.33

1.36

1.0

1.8

1.0

3

0.7

1.71

43

70

76

1.05

1.01

1.02

1.0

2.0

1.0

4

1.0

1.12

60

106

128

1.29

1.26

1.31

1.0

2.3

1.0

5

0.7

1.28

28

121

143

1.28

1.68

1.38

1.0

1.1

1.0

6

1.2

2.00

31

245

251

1.45

1.08

1.20

1.0

1.1

1.0

7

0.6

2.50

24

160

170

1.07

1.18

1.07

1.0

1.2

1.0

8

0.9

4.66

23

112

120

1.14

1.12

1.14

1.0

1.3

1.0

9

0.7

13.30

28

80

82

1.14

1.25

1.14

1.0

1.0

1.0

10

0.4

10.76

29

97

99

1.06

1.08

1.12

1.0

1.0

1.0

11

0.4

29.09

28

94

101

1.14

1.20

1.22

1.0

1.0

1.0

12

0.4

9.37

29

80

82

1.05

1.08

1.10

1.0

1.0

1.0

13

0.8

7.62

27

81

84

1.15

1.22

1.18

1.0

1.0

1.0

10.2

4.86

36

107

118

-

-

-

-

-

-

FIGURE 6

A positive yet not statistically significant relationship relates the change rates of channel width with sinuosity. Larger changes occurred between 2005 and 2009 rather than from 2009 to 2012 (Fig. 7).

FIGURE 7

Large wood: The total amount of LW pieces within the active channel along the whole study segment increased significantly (P <0.001) from 16 LW pieces in 2005 to 756 in 2009 (Table 3). Comparing data of year 2012 with 2009, no statistically significant differences were found in the number of LW pieces and number of log jams in the channel.

Between 2009 and 2012 the number of LW pieces increased in six reaches (major increase was 270%) and decreased in seven (major decrease 34%). However, if the entire channel segment is considered, the number of wood pieces decreased by 2.6% (from 756 to 736). During the same period the number of log jams increased in five reaches (higher increase of 210%) and decreased in six (major decrease up to 85%), while a net increase of 4% (from 302 to 314) was measured in the whole segment (Table 3).

Table 3. Number of individual wood pieces and log jams for each reach and year, and variation between 2009 and 2012.

Reach 

N Individual wood pieces

Percent change, 2009 to 2012

N Log jams

Log jams variation between 2009 and 2012

2005

2009

2012

Variation (%)

2005

2009

2012

Variation (%)

1

0

28

25

-10

0

13

26

+100

2

12

81

88

+9

0

30

93

+210

3

2

11

41

+273

0

8

16

+100

4

2

95

116

+22

0

65

61

-6

5

0

86

112

+30

0

74

47

-36

6

0

159

111

-30

0

68

44

-35

7

0

52

35

-33

0

13

2

-85

8

0

105

69

-34

0

10

4

-60

9

0

21

34

+62

0

2

3

+50

10

0

8

7

-13

0

0

0

-

11

0

19

14

-26

0

2

3

+50

12

0

24

29

+21

0

2

2

0

13

0

67

55

-18

0

15

13

-13

Total 

16

756

736

-3

0

302

314

4

At the reach level, the mean number of large wood pieces (N 100 m-1) was 0.2 in 2005, 7.4 in 2009 and 7.0 in 2012, and the number of log jams (N km-1) were on average 3 for 2009 and 2012, respectively (Fig. 8).

LW abundance either as individual pieces or in log jams were not related with the catchment area.

The mean length of the individual wood pieces was 8.2 and 10 m in 2009 and 2012, respectively, and this difference is statistically significant. However it is worth pointing out that this difference is very close to the errors associated with the measurements of log pieces from the aerial photos. The mean log jam size was 156 and 201 m2 in 2009 and 2012, respectively, even if this difference resulted to be not statistically significant.

FIGURE 8

Overall, steeper reaches featured lower abundance of LW, and logs were mainly accumulated in log jams. Fig. 9 shows the distribution of LW along the study segment, expressed it in number of individual pieces of large wood and number of log jams per reach active channel length and area for both years 2009 and 2012.

FIGURE 9

Considering the full segment the lengths of pieces of large wood were 8.3 (±3.9) m and 10 (±4) m for 2009 and 2012 respectively. On the other hand, mean area of the segment logjams was 156 and 201 m2 for 2009 and 2012, respectively, with minimum and maximum mean reach areas of 10 and 9,740 m2 for 2009 and 12 and 5,609 m2 for 2012.

Large woody and channel morphology: Table 4 shows the relations among LW distribution characteristics as dependent variables and channel morphology features as independent variables, the R2 coefficients and level of statistical significance among variables. Data for the 2005 condition is not included in Table 4 as LW appears in only three reaches. Some of the statistically significant relationships are shown in Fig. 10 along with the regressions. Overall, the longitudinal distribution of LW is positively related with channel sinuosity (Table 4). This tendency was statistically significant in years 2009 and 2012 with respect to distribution of LW jams (Fig. 10a) and individual LW pieces (N km-1, Fig. 10b), and only for year 2009 in relation to the surface of log jams per channel unit area.

It was observed that channel width does not statistically explain the distribution of LW jams (Table 4). However, this variable shows a positive trend with respect to the number of large wood pieces, and a statistically significant relationship only with the number of LW pieces per kilometer of channel length (N km-1, Fig. 10d). Regarding the surface of log jams by reach, a positive and statistically significant relation was observed with channel width only for the 2009 condition. Finally, the maximum size of jams per reach shows a statistically significant relation with channel width both in 2009 and 2012.

Table 4. Relations between variables of LW distribution by reach and morphologic channel variables.

 

 

 

Sinuosity

Channel width (m)

2009

2012

2009

2012

Number of Log jams

Nº Km-1

2009

y = 71.2e-0.33x R2=0.63; P<0.01

R2=0.19; ns

2012

y = 71.2e-0.33x R2=0.48; P<0.01

R2=0.10; ns

Nº ha-1

2009

y = 10.9x - 10.4 R2=0.54; P<0.01

R2=0.05; ns

2012

 

y = 7.07x - 5.89 R2=0.22; P<0.1

 

R2=0.00; ns

Number of LW

Nº Km-1

2009

y = 170.9x - 128 R2=0.36; P<0.05

 

y = 0.58x + 3.6 R2=0.52; P<0.01

 

2012

y = 215x - 185.4 R2=0.47; P<0.05

y = 0.39x + 22.3 R2=0.28; P<0.1

Nº ha-1

2009

R2=0.54; ns

R2=0.07; ns

2012

 

R2=0.22; ns

 

R2=0.00; ns

Surface of Log jams

ha ha-1

2009

y = 0.13x - 0.12 R2=0.42; P<0.05

 

y = 0.0003x - 0.001 R2=0.32; P<0.05

 

2012

 

R²= 0.19; ns

 

R2=0.04; ns

Jams maximum size

ha

2009

R²= 0.20; ns

 

y = 0.0045x - 0.38 R2=0.70; P<0.01

 

2012

 

R²= 0.20; ns

 

y = 0.0023x - 0.2 R2=0.66; P<0.01

A positive relation was found between the number of LW jams (N km-1 and N ha-1) and the active channel incremental area in years 2009 and 2012. The relation is statistically significant with LW jams in N km-1 and N ha-1 for 2012 and only in N ha-1 in 2009 (Fig. 10c). No relation was found between the spatial distribution of individual LW pieces with accumulated channel area. Only in 2009, a statistically significant relationship was found between reach jams surface and the accumulated channel area, while no relation was found with maximum size of jams.

FIGURE 10

At the reach level, the differences in number of individual pieces of large wood and logjams between the two post-eruption periods were not statistically correlated with channel morphologic features such as channel width, sinuosity and slope.

Supervised classification: Fig. 11 shows two channel reaches where large wood were recognised using the manual digitalization (b) and (e) and using the supervised classification process (c) and (f). Overall, both methodologies provide very similar results in terms of proportion and LW distribution.

FIGURE 11

Considering the total of individual pieces of large wood and log jams, in the reach with lower presence of wood material, 66 and 849 N km-1 were identified with the traditional and supervised classification methods, respectively. In the reach with high LW abundance, 213 and 2604 individual pieces of large wood and log jams (in N km-1) were recognized using the traditional method and the supervised classification, respectively. The number of large wood (N km-1) documented in the reach with low presence of wood material were 31 and 33% of LW identified in the reach with high LW abundance, with the traditional and supervised classification techniques, respectively.

5 Discussions

Morphologic changes en el ancho, sinuosidad y forma (braiding): Evidence provided show that the morphology of the Blanco River changed dramatically after the Chaitén Volcano eruption, confirming the observations of Meyer and Martinson (1989), Major (2003) and Pierson and Major (2014), among others, on the magnitude of impacts on river systems affected by volcanic eruptions. Pyroclastic flows, lahars and landslides caused by the eruption of the Chaitén Volcano were quickly deposited along the channel and over the stream banks of the Blanco River. Flows with high velocities and sediment load originated from the remobilization of the ash cover and the further collapse of the dome of the Chaitén Volcano (Major et al. 2013) favored wood entrainment and transport (Swanson et al. 2013) removing the riparian forests, aggrading the river up to eight meters, and causing lateral channel change. The analysis of the sequence of remote sensing imagery allowed to verify that the channel width increased in average by 3.5 times when comparing the 2009 condition to the original pre-eruption situation. The maximum channel enlargement occurred in the middle of the study segment (reaches 5 to 8) probably because of a reduction of the longitudinal slope and the widening of the valley, which likely reduced the speed of the pyroclastic flows. Major et al. (2013) indicated that the new pyroclastic flow originated in February 2009 moved up to seven kilometers downstream from the volcano dome. Reaches 5 to 8 are located at a distance between 3 and 6.5 kilometers from the volcano and in this sector channel enlargements reached up to 9 times if compared to the original pre-eruption width. Increases of channel width as observed in the Blanco River are of the same order of magnitude than those previously observed after volcanic eruptions such as Mount St. Helens (Meyer and Martinson 1989) and Mount Pinatubo (Hayes et al. 2002).

Between the two post-eruption periods (2009 and 2012) some cross sections showed changes in channel widths higher than the 100%, but these changes were highly site-specific. However, changes in reach channel widths among these two periods were statistically non-significant.

Higher variations in the sinuosity and braiding indices were observed between 2009 and the pre-eruption condition than from 2012 and 2009, indicating that the most dramatic channel morphologic changes occurred within the first years after the eruption. The lower rate of change from 2009 to 2012 seems to suggest that the Blanco River is in a recovery phase, reaching in some areas its original channel, in agreement with what observed in comparable studies (e.g. Meyer and Martinson 1989). Similar trends appear when comparing the changes in channel width over the post-eruption years. For instance, channel width remained fairly constant from 2009 to 2012 probably because of the lack of new lahars and large flows.

Sinuosity index and channel width (see Fig. 6) are significantly related when excluding data from reaches 6 and 7 (outliers inside a circle in Fig. 6). These outliers could be explained by the combined effects between pyroclastic flows and changes in the longitudinal slope along the study segment and the width of the floodplain in reaches downstream the confluence of the Caldera Creek. In fact, in reaches where the longitudinal slope starts to decrease and floodplain is wider (i.e. where volcanic flows reduced their speed), the highest rates of change of the channel width and sinuosity occurred. This can be seen in Fig. 12 which shows that reaches 5 to 8 concentrate most of the changes of width and sinuosity.

FIGURE 12

Before the eruption, the study segment featured a single channel. The 2009 image shows that between reaches 2 to 8 there were more than one channel and reaches 2 to 4 could be classified as braided. However, analysis of the 2012 image shows that in the study segment the Blanco River returned to a single channel pattern (see Table 2), although the 2012 channel was yet not recovered as before the eruption.

Large wood: As a result of the destruction of the vegetation within the whole fluvial corridor by pyroclastic flows and lahars (Swanson et al. 2013), the Blanco River channel experienced a massive supply of dead wood (see Table 3 and Fig. 8). These increases of large wood in channels are explained due to the ability of these eruptions to destroy vegetation, as reported by Meyer and Martinson (1989) and Lisle (1995). These authors note that wood in streams is relevant in controlling erosion and post-eruption channel adjustment.

Abundant wood material can be observed throughout the study segment, but the rate of large wood accumulation decreases in the reaches closer to the volcano as shown in Fig. 9. This can be explained by the higher slope of these reaches which undoubtedly increased stream power and large wood transport capacity, mostly because the steep bed slopes determine frequent debris flows able to flush the stored logs (Rigon et al. 2012). On the other hand, post-eruption channel width in the Caldera Creek (minimum width of 31 m) is wide enough for wood transport, and therefore it is expected that sections with higher slope would feature lower number of individual LW and log jams. In the downstream reaches, the lower slopes and the larger channel width reduce available energy for LW transport favoring the deposition of wood material.

Interestingly, no obvious relationship between large wood storage and temporal variation of large wood was found, i.e reaches with higher wood volume did not necessarily show higher temporal variation in wood abundance or storage than in reaches with less volume.

Large wood and morphology: Overall, after the eruption there was a huge recruitment of large wood material along the study segment. Lisle (1995) also report abundance of wood material in channels affected by pyroclastic flows at Mount Saint Helens. However, the distribution of large wood along the study channel appears heterogeneous and the accumulation of both individual pieces of large wood and log jams is significantly related with the channel sinuosity, width and cumulative area. As previously reported by Moulin et al. (2011), wood jams generally tend to be found in reaches with highest sinuosity (Fig. 10), probably because of the higher roughness of the banks. There is also a positive relationship between the number of log jams and the surface of accumulations with the cumulative area of the channel, probably due to the downstream movement and subsequent deposit. Reach scale abundance of large wood is not related to the drainage area at the upstream end of each reach, likelly because the study segment features two distinctivly different zones: the lower part with a flatbed channel of the main Blanco River draining a larger basin, and the upper and steeper part, the Caldera Creek, draining a much smaller basin. It should also be considered that the distribution of LW is not only influenced by the local morphologic characteristics of the channel, but by the characteristics of pyroclastic flows and lahars as well.

Several channel morphological parameters are significantly related with distribution features of wood material (see Table 4). If individual wood pieces are considered, LW distribution (in N km-1) is significantly correlated with channel sinuosity and width (Table 4 and Fig. 10), probably because sinuous or wider reaches allow wood recruitment of elements traveling from upstream reaches. It is worth mentioning here that Iroumé et al. (2011), in a study in the Chilean Cordillera basins, found that bankfull width is the only channel parameter having a significant relationship with the longitudinal distribution of wood pieces. Relationships show that the amount of wood material increases as channel widen as previously showed by Fox and Bolton (2007), who found greater amount of wood in wider channels or larger watersheds. However, it should be stressed what reported by Fox and Bolton (2007) is quite a unique case, as most previous investigations found that this relationship is inverse, i.e. the larger the basin, the less amount of LW (Marcus et al. 2002, Abbe and Montgomery 2003, Wohl and Jaeger 2009, Wohl and Cadol 2011, Rigon et al. 2012). However, it is important to note that the current investigation deals with the uncommon situation of severe impact generated by a volcanic eruption. Most of the channel reaches showed important changes in the number of individual pieces of large wood or logjams (Table 3), but these variations are not statistically correlated with channel morphologic variables. In the Blanco River the number of individual pieces of large wood elements and log jams did not change between 2009 and 2012. However, a comparison of the 2009 and 2012 images show that only 17% of the individual wood had remained in place and 78% had been displaced, while it was not possible to clearly define the final location for the remaining 5%. Forty-eight percent of wood jams was removed, while 47% remained in place between 2009 and 2012; the rest of the accumulations remained in place but their size was modified. The mean length of the mobilized LW pieces was significantly lower than those who remain in the same location (7.9 and 9.5 m, respectively). At the same time, the mean area of log jams remaining in place is significantly higher than those who were displaced (167 and 78 m² respectively). These differences seem reasonable as shorter LW pieces and smaller log jams should be more easily prone to be moved. In spite of high LW movement and variation rates, no significant changes were observed during the 2009-2012 period as to the amount or distribution of wood material along the channel, probably because of the almost unlimited LW supply capacity from the channel margins and due to stream bank erosion and channel incision which expose buried wood material.

Even if the displacement length of single logs could not be monitored, it seems reasonable to assume that part of the wood pieces transported to the downstream end of the Blanco River were deposited near of Chaitén Town and collected by local inhabitants, while the rest were deposited into the delta. Because no artificial or natural obstacles are locatedalong the channel, the existing logjams were created naturally by wood deposits on bars and channel margins.

Supervised classification: Large wood data collection has been traditionally carried out in the field, with obvious restrictions on the extent of survivable area and chances of repeting the survey overtime (Leckie et al. 2005). The use of remote imagery for LW recognition represented a significant step forward to overcoming these time and cost restrictions. Indeed, the use of a supervised classification technique to identify wood elements provides valuable results with respect to the location of material, identification of areas with higher densities of this material, and comparison of quantity of wood material per unit of area for different channel reaches. However, severe restrictions seem to remain, especially due to the misinterpretation of pixels by the algorithms, mainly in shadow areas which are common in mountain and complex terrains. Another source of error clearly arising from the interpretation of the images from the Blanco River was due to ashes deposited on the large wood pieces which made difficult to distinguish between sediment and wood. Further limitations are due to the need of a post-supervised classification process to reduce errors in the interpretation. This requires additional time, and for the interpretation of limited surfaces this could match or exceed the time required for traditional manual image interpretation, especially if high precision levels of interpretation are required. Similar restrictions were observed by Leckie et al. (2005) in the interpretation of imagery using supervised classification. However, the use of the supervised classification technique has advantages to study large wood abundance and distribution using high spatial resolution imagery in open areas free from vegetation as the case of the Blanco River segment, as noted by Charlton et al. (2003).

6 Conclusions

When compared with the original pre-eruption condition, statistically significant changes were identified in the channel morphology and large wood abundance and distribution on the Blanco River study segment after the eruption of the Chaitén Volcano. Channel width increased by 211% and total wood pieces from 16 to 736 pieces of large wood between 2005 and 2009. No significant changes were observed in channel width and large wood abundance during the period 2009-2012.

The shape of channel was strongly affected by the volcanic eruption, mainly due to pyroclastic flows, landslides, mudflows and heavy rainfall that caused flooding. The greatest changes in the channel type occurred in the downstream low-slope reaches of the study segment. In 2009, greater sinuosity was observed in several reaches with three of them having a braided pattern. Channel sinuosity changed from 2005 and 2009, and by 2012 the channel tended to recover its original pre-eruption condition.

Several variables of large wood distribution in the study segment have clear and statistically significant trends when associated with channel morphologic conditions. For instance, the sinuosity is positively and significantly related to the number of wood jams per channel kilometer or channel hectare. Besides, the maximum areal size of accumulations has a positive and statistically significant relationship with the channel width.

Overall, the greatest river changes in terms of channel morphology and LW amount occurred between the eruption and 2009, whereas from 2009 to 2012 channel morphology tends to return to pre-eruption values due to incision on the loose sediment deposits. On the other hand, even if the mobility of LW is remarkably high (78 % and 48 % of mobility rate for single logs and jams, respectively) the amount of LW within the channel is still much higher than the pre-eruption condition.

This study highlights the importance of monitoring morphological evolution, sediment transport and large wood dynamics in river systems affected by paroxistic events such as eruptions. Although differences among reaches and high mobility, high volumes of LW still remain in the channel bed and margins. This LW availability and the observations of focalized but important changes in channel widths indicate that the Blanco River will continue adjusting for a much longer period than the one considered in this research. Further investigation on this study site will allow monitoring the rates of channel incision over pyroclastic deposits until reaching the original channel bed and the dynamics of large wood that will be supplied from buried bank deposit of logs due to lateral channel erosion.

Acknowledgements

This study is part of a doctoral thesis in Forest Sciences, Universidad Austral de Chile, and is supported by Project FONDECYT 1110609.

References

Alfano, F., Bonadonna, C., Volentik, A.C.M., Connor, C.B., Watt, S.F.L., Pyle, D.M., Connor, L.J., 2011. Tephra stratigraphy and eruptive volume of the May, 2008, Chaitén eruption, Chile. Bulletin of Volcanology, 73(5), 613-630

Andreoli, A., Comiti, F. and Lenzi, M.A., 2007a. Characteristics, distribution and geomorphic role of large woody debris in a mountain stream of the Chilean Andes. Earth Surface Processes and Landforms, 32, 1675–1692.

Andreoli, A., Carlig, G., Comiti, F. and Iroumé, A., 2007b. Residuos leñosos de gran tamaño en un torrente de la Cordillera de los Andes, Chile: su funcionalidad e importancia. Bosque, 28(2), 83-96.

Andreoli, A., Comiti, F., Mao, L., Iroumé, A. and Lenzi, M.A., 2008. Evaluación de los volúmenes y de los efectos hidro-morfológicos del material leñoso en dos torrentes andinos (Chile). Ingeniería del Agua, 15(3), 189-204.

Bertoldi, W., Gurnell, A.M. and Drake, N.A., 2011. The topographic signature of vegetation development along a braided river: Results of a combined analysis of airborne LiDAR, color air photographs, and ground measurements. Water Resources Research, 47, W06525. doi:10.1029/2010WR010319.

Branca, S., 2003. Geological and geomorphological evolution of the Etna volcano NE flank and relationships between lava flow invasions and erosional processes in the Alcantara Valley (Italy). Geomorphology, 53(3), 247-261.

Buckley, A.R., Fleece, W. and Renslow, M., 2000. Development of stream indicators using LIDAR (Light Detection and Ranging) data GIS/EM4 No. 206. 4th International Conference on Integrating GIS and Environmental Modeling (GIS/EM4): Problems, Prospects and Research Needs. September 2 - 8, 2000, Banff, Alberta, Canada.

Bull, J.M., Miller, H., Gravley, D., Costello, D., Hikuroa, D. and Dix, J., 2010. Assessing debris flows using LIDAR differencing: 18 May 2005 Matata event, New Zealand. Geomorphology, 124, 75–84.

Cavalli, M., Tarolli, P., Marchi, L. and Dalla Fontana, G., 2008. The effectiveness of airborne LiDAR data in the recognition of channel-bed morphology. Catena, 73(3), 249-260.

Carn, S.A., Pallister, J.S., Lara, L., Ewert, J.W., Watt, S., Prata, A.J., Thomas, R.J. and Villarosa, G., 2009. The unexpected awakening of Chaitén Volcano, Chile. Eos, Transactions, American Geophysical Union, 90(24), 205–206.

Charlton, M.E., Large, A.R. and Fuller, I.C., 2003. Application of airborne LiDAR in river environments: the river Coquet, Northumberland, UK. Earth Surface Processes and Landforms, 28, 299-306.

Cole, P.E., Calder, E.S., Sparks, R.S.J., Clarke, A.B., Druitt, T.H., Young, S.R., Herd, R., Harford, C.L. and Norton, G.E., 2002. Deposits from dome-collapse and fountain-collapse pyroclastic flows at Soufrière Hills Volcano, Montserrat, in The Eruption of the Soufrière Hills Volcano, Montserrat From 1995 to 1999, edited by T. H. Druitt and B. P. Kokelaar. Geological Society London Memoirs, 21, 231– 262.

Comiti, F., Andreoli, A., Mao, L. And Lenzi, M.A., 2008a. Wood storage in three mountain streams of the Southern Andes and its hydro-morphological effects. Earth Surface Processes and Landforms, 33, 244–262.

Comiti, F., Pecorari, E., Mao, L., Picco, L., Rigon, E. and Lenzi, M.A., 2008b. New Methods for Determining Wood Storage and Mobility in Large Gravel-bed Rivers. EPIC FORCE project report (D20bis) University of Padova, Padova, Italy.

CONAF. 1997. Catastro y evaluación de los recursos vegetacionales nativos de Chile. Universidad Austral de Chile, Pontificia Universidad Católica de Chile y Universidad Católica de Temuco.

Dale, V.H., Swanson, F.J. and Crisafulli, C.M., 2005. Ecological responses to the 1980 eruption of Mount St. Helens. Springer-Verlag, New York.

Diaz, E. and Ollero, A., 2005. Metodología para la clasificación geomorfológica de los cursos fluviales de la cuenca del Ebro. Geographicalia, 47, 23-45.

Digimapas-Chile, 2009. Informe final: Proyecto "Servicio de levantamiento aerofotogramétrico y MDE con láser aerotransportado para el sector de Santa Barbara, comuna de Chaiten”. Santiago, Chile, 15 pp.

Donoso, C., 1981. Tipos forestales de los bosques nativos de Chile. Doc. De trabajo N° 38. Proyecto FAO. FO: DP/CHI/76/003.

Fox, M. and Bolton, S., 2007. A regional and geomorphic reference for quantities and volumes of instream wood in unmanaged forested basins of Washington State. North American Journal of Fisheries Management, 27(1), 342-359.

Gran, K.B. and Montgomery, D.R., 2005. Spatial and temporal patterns in fluvial recovery following volcanic eruptions: channel response to basin-wide sediment loading at Mount Pinatubo, Philippines. Geological Society of America Bulletin, 117, 195–211.

Gurnell, A.M., Piegay, H., Swanson, F.J. and Gregory, S.V., 2002. Large wood and fluvial processes. Freshwater Biology, 47, 601–619.

Harpel, C.J., de Silva, S., and Salas, G., 2011. The 2 ka Eruption of Misti Volcano, Southern Peru—The Most Recent Plinian Eruption of Arequipa's Iconic Volcano. Geological Society of America Special Papers, 484, 1-72.

Hayes, S.K., Montgomery, D.R. and Newhall, C.G., 2002. Fluvial sediment transport and deposition following the 1991 eruption of Mount Pinatubo. Geomorphology, 45, 211–224.

Horacio, J. and Ollero, A., 2011. Clasificación geomorfológica de cursos fluviales a partir de Sistemas de Información Geográfica (S.I.G.). Boletín de la Asociación de Geógrafos Españoles, 56.

Inbar, M., Reyes, A. and Graniel, J.H., 2001. Morphological changes and erosion processes following the 1982 eruption of El Chichón volcano, Chiapas, Mexico. Géomorphologie: relief, processus, environnement, 7(3), 175-183.

Iroumé, A., Ulloa, H., Lenzi, M.A., Andreoli, A. and Gallo, C., 2011. In-stream large wood mobility and recruitment in two channels in the Coastal Mountain Range, Chile. Bosque, 32(3), 247-254.

Iroumé, A., Mao, L., Ulloa, H., Ruz, C. and Andreoli, A., 2014. Large wood volume and longitudinal distribution in channel segments draining catchments with different land use, Chile. Open Journal of Modern Hydrology, 4, 57-66.

Lara, L., 2009., The 2008 eruption of the Chaitén Volcano, Chile: A preliminary report. Andean Geology, 36(1), 125–129.

Leckie, D.G., Cloney, E., Jay, C. and Paradine, D., 2005. Automated Mapping of Stream Features with High-Resolution Multispectral Imagery: An Example of the Capabilities. Photogrammetric Engineering & Remote Sensing, 71(2), 145–155.

Lisle, T.E., 1995. Effects of coarse woody debris and its removal on a channel affected by the 1980 eruption of Mt. St. Helens,Washington. Water Resources Research, 31, 1797–1808.

Major, J., Pierson, T.C., Dinehart, R.L. JE Costa, J.E., 2000. Sediment yield following severe volcanic disturbance-A two decade perspective from Mount St. Helens. Geology, 28, 819–822.

Major, J., 2003. Post-eruption hydrology and sediment transport in volcanic river systems. Water Resources Impact, 5(3), 10-15.

Major, J., 2004. Posteruption suspended sediment transport at Mount St. Helens: Decadal-scale relationships with landscape adjustments and river discharges. Journal of Geophysical Research, 109, F01002. doi:10.1020/2002JF000010

Major, J.J., Crisafulli, C.M., Frenzen, P., and Bishop, J., 2009. After the disaster: The hydrogeomorphic, ecological, and biological responses to the 1980 eruption of Mount St. Helens, Washington. Field Guides, 15, 111-134.

Major, J. J. and Lara, L. E. 2013. Overview of Chaitén Volcano, Chile, and its 2008-2009 eruption. Andean Geology, 40(2), 196-215.

Major, J.J., Pierson, T.C., Hoblitt, R.P., and Moreno, H., 2013, Pyroclastic density currents associated with the 2008-09 eruption of Chaitén Volcano (Chile)—forest disturbances, deposits, and dynamics. Andean Geology, 40(2), 324-358.

Manville, V., 2002. Sedimentary and geomorphic responses to ignimbrite emplacement: Readjustment of the Waikato River after the A.D. 181 Taupo eruption, New Zealand. The Journal of Geology, 110, 519– 541.

Manville, V. and Wilson, C.J.N., 2004. The 26.5 ka Oruanui eruption, New Zealand: a review of the roles of volcanism and climate in the post-eruptive sedimentary response. New Zealand Journal of Geology and Geophysics, 47, 525–547.

Manville, V., Newton, E. H., and White, J. D., 2005. Fluvial responses to volcanism: resedimentation of the 1800a Taupo ignimbrite eruption in the Rangitaiki River catchment, North Island, New Zealand. Geomorphology, 65(1), 49-70.

Mao, L., Burns, S., Comiti, F., Andreoli, A., Urciuolo, A., Gaviño-Novillo, M., Iturraspe, R. and Lenzi, M.A., 2008a. Acumulaciones de detritos leñosos en un cauce de montaña de Tierra del Fuego: análisis de la movilidad y de los efectos hidromorfológicos. Bosque, 29 (3), 197-211.

Mao, L., Andreoli, A., Comiti, F. And Lenzi, M.A., 2008b. Geomorphic effects of large wood jams on a sub-Antarctic mountain stream. River Research and Applications, 24, 249–266.

Marcus, W.A., Marston, R.A., Colvard Jr, C.R. and Gray, R.D., 2002 Mapping the spatial and temporal distributions of large woody debris in rivers of the Greater Yellowstone Ecosystem, USA. Geomorphology, 44, 323-335.

Meyer, D.F. and Martinson, H.A., 1989. Rates and Processes of Channel Development and Recovery Following the 1980 Eruption of Mount St. Helens, Washington. Hydrological Sciences Journal, 34, 115-127.

Moody, J.A., and Martin, D.A., 2001. Initial hydrologic and geomorphic response following a wildfire in the Colorado Front Range. Earth Surface Processes and Landforms, 26, 1049–1070.

Moulin, B., Schenk, E. R., Hupp, C. R., 2011. Distribution and characterization of in‐channel large wood in relation to geomorphic patterns on a low‐gradient river. Earth Surface Processes and Landforms, 36(9), 1137-1151.

Oppenheimer, C., 2003. Climatic, environmental and human consequences of the largest known historic eruption: Tambora volcano (Indonesia) 1815. Progress in physical geography, 27(2), 230-259.

Pallister, J.S., Major, J., Pierson, T. and Hoblitt, R., 2010. Interdisciplinary Studies of Eruption at Chaitén Volcano, Chile. Eos, 91, 42-19.

Pallister, J.S., Diefenbach, A.K., Burton, W.C., Muñoz, J., Griswold, J.P., Lara, L.E., Lowenstern, J.B., and Valenzuela, C. E., 2013. The Chaitén rhyolite lava dome: Eruption sequence, lava dome volumes, rapid effusion rates and source of the rhyolite magma. Andean Geology, 40(2), 277-294.

Peralta M. 1980. Consideraciones generales para el uso de los suelos, principalmente forestales, de la región de Alto Palena y Chaiten X region. Boletin Tecnico, Facultad de Ciencias Forestales, Universidad de Chile, Santiago 58, 1–49.

Piegay, H. and Marston, R.A., 1998. Distribution of large woody debris along the outer bend of meanders in the Ain River, France. Physical Geography, 19, 318– 340.

Pierson, T., Pringle, P., and Cameron, K., 2011. Magnitude and timing of downstream channel aggradation and degradation in response to a dome-building eruption at Mount Hood, Oregon. Geological Society of America Bulletin, 123(1-2), 3-20.

Pierson, T., Major, J., Amigo, A., and Moreno, H., 2013. Acute sedimentation response to rainfall following the explosive phase of the 2008–09 eruption of Chaitén Volcano, Chile. Bulletin of Volcanology, 75(5), 1-17.

Pierson, T., and Major, J., 2014. Hydrogeomorphic Effects of Explosive Volcanic Eruptions on Drainage Basins. Annu. Rev. Earth Planet. Sci., 42, 469–507.

Piña-Gauthier, M., Lara, L.E., Bataille, K., Tassara, A., Baez, J.C. 2013. Co-eruptive deformation and dome growth during the 2008-2009 Chaitén eruption, Southern Andes. Andean Geology, 40 (2), 301-323.

Rigon, E., Comiti, F., and Lenzi, M.A., 2012. Large wood storage in streams of the Eastern Italian Alps and the relevance of hillslope processes. Water Resources Research, 48(1).

Rinaldi, M., Surian, N., Comiti, F. and Bussettini, M., 2010. Sistema di Valutazione Morfologica dei corsi d’acqua – Manuale tecnico-operativo per la valutazione ed il monitoraggio dello stato morfo-logico dei corsi d’acqua - Versione 0. Istituto Superiore per la Protezione e la Ricerca Ambientale, Roma, 191 pp. - ISBN: 978-88-448-0438-1.

Rosgen, D.L., 1994. A classification of natural rivers. Catena, 22, 169-199.

Smikrud, K., Prakash, A. and Nichols, J., 2008. Decision-Based Fusion for Improved Fluvial Landscape Classification Using Digital Aerial Photographs and Forward Looking Infrared Images. Photogrammetric Engineering & Remote Sensing, 74(7), 903–911.

Snyder, N.P., 2009. Studying Stream Morphology with Airborne Laser Elevation Data. Eos, 90(6), 45–52.

Swanson, F.J., Jones, J.A., Crisafulli, C., Lara, A. 2013. Effects of volcanic and hydrologic processes on forest vegetation, Chaitén Volcano, Chile. Andean Geology 40 (2): 359-391.

Tanarro, L.M., Andrés, N., Zamorano, J.J., Palacios, D., and Renschler, C.S., 2010. Geomorphological evolution of a fluvial channel after primary lahar deposition: Huiloac Gorge, Popocatépetl volcano (Mexico). Geomorphology, 122(1), 178-190.

Ulloa, H., Iroumé, A., Lenzi, M.A., Andreoli, A., Álvarez, C. and Barrera, V., 2011. Large wood in two catchments from the Coastal Mountain range with different land use history. Bosque, 32(3), 235-245.

Wisner, B., Blaikie, P., Cannon, T. and Davis, I., 2003. At risk: natural hazards, people's vulnerability and disasters. Second edition. London, Inglaterra. Routledge. 492 pp

Wohl, E., and Jaeger, K., 2009. A conceptual model for the longitudinal distribution of wood in mountain streams. Earth Surface Processes and Landforms, 34, 329–344.

Wohl, E., and Cadol, D., 2011. Neighborhood matters: Patterns and controls on wood distribution in old-growth forest streams of the Colorado Front Range, USA. Geomorphology, 125(1), 132-146.

Woll, C., Prakash, A. and Sutton, T., 2011. A Case-Study of in-Stream Juvenile Salmon Habitat Classification Using Decision-Based Fusion of Multispectral Aerial Images. Applied Remote Sensing Journal, 2(1), 37-46.

Zheng, S., Wu, B., Thorne, C., and Simon, A., 2014. Morphological evolution of the North Fork Toutle River following the eruption of Mount St. Helens, Washington. Geomorphology. 208, 102–116

2

02468101.01.21.41.6

Width 09 / Width 05Sinuosity 2005

R

2

= 0.22; ns

0.00.40.81.21.61.01.21.41.6

Width 12 / Width 09Sinuosity 2009

R

2

= 0.21; ns(a)(b)

(a)(b)

0246812345678910111213

NLog jams ha

-1

02468101212345678910111213

NLW ha

-1

05010015020012345678910111213

NLW km

-1

2009201202040608010012012345678910111213

NLog jams km

-1

20092012

Reach

(a)(b)(c)(d)

0501001502000100200300

Nº LW km

-1

Bankfull width

0204060801001201.01.52.0

NLog jamskm

-1

0501001502001.01.52.0

NLW km

-1

020406080100120050100150

Nº Log jams km

-1

Bankfull acumulatedarea (Ha)

R² = 0.63;P< 0.01R² = 0.48; P< 0.01 R² = 0.36;P< 0.05 R² = 0.47; P< 0.05 R² = 0.20;nsR² = 0.57; P< 0.01 R² = 0.52;P< 0.01 R² = 0.28; P< 0.1

Sinuosity

(c)(b)(d)(a)

200520092012200920052012Channel area (ha km

-1

)Channel width (m)

(a)(b)

R² = 0.2059

0.00.40.81.21.60100200300

Channel width (m)R² = 0.1316

0.00.40.81.21.60100200300

R² = 0.0648

0.00.40.81.21.60204060

Sinuosity (a)(b)(c)R

2

= 0.06; nsR

2

= 0.21; nsR

2

= 0.13; ns