Assessment of the debris-flow susceptibility in tropical mountains using clast distribution patterns

Assessment of the debris-flow susceptibility in tropical mountains using clast distribution patterns

    Assessment of the debris-flow susceptibility in tropical mountains using clast distribution patterns La´ıs de Carvalho Faria Lima Lop...

1MB Sizes 11 Downloads 39 Views

    Assessment of the debris-flow susceptibility in tropical mountains using clast distribution patterns La´ıs de Carvalho Faria Lima Lopes, Lu´ıs de Almeida Prado Bacellar, Paulo de Tarso Amorim Castro PII: DOI: Reference:

S0169-555X(16)30195-7 doi: 10.1016/j.geomorph.2016.09.026 GEOMOR 5773

To appear in:

Geomorphology

Received date: Revised date: Accepted date:

15 April 2016 22 August 2016 12 September 2016

Please cite this article as: de Carvalho Faria Lima Lopes, La´ıs, de Almeida Prado Bacellar, Lu´ıs, de Tarso Amorim Castro, Paulo, Assessment of the debris-flow susceptibility in tropical mountains using clast distribution patterns, Geomorphology (2016), doi: 10.1016/j.geomorph.2016.09.026

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Assessment of the debris-flow susceptibility in tropical mountains using clast distribution patterns.

T

Laís de Carvalho Faria Lima Lopes a, Luís de Almeida Prado Bacellar ab, Paulo de Tarso

IP

Amorim Castro a

SC R

Email adresses: [email protected] (L. Lopes); [email protected] (L. Bacellar); [email protected] (P. Castro) a

Ouro Preto Federal University, Campus Morro do Cruzeiro, 35400-000, Ouro Preto (MG), Brazil.

b

Corresponding author – Address: UFOP/DEGEO – Campus Morro do Cruzeiro s/n – CEP 35400-000 -

MA

NU

Bauxita – Ouro Preto – MG – Brazil; Phone: 55 31 991455360.

ABSTRACT

D

Channel morphometric parameters and clast distribution patterns in selected basins of the Ferriferous Quadrangle tropical mountains, Brazil, were analyzed in order to assess

TE

susceptibility to debris flows. Median bed surface clast size (D50) in the main stream channel of these basins shows a coarsening downstream trend with drainage areas of up to 6 km2,

CE P

which is attributed to debris flow dominated-channels by some authors. The composition and roundness of the bed load, clast sand, and the presence of allochthonous large boulders throughout the channels also suggest the occurrence of past debris flow in the region.

AC

Luminescence Optically Stimulated (LOE) dating points out that debris flow could have occurred as a consequence of climate changes in the Late Pleistocene and Holocene and it can now be triggered by deforestation or extreme rainfall events. There has not been any record of past debris flow in the study area, or in other mountainous regions of Brazil where debris flows have recently occurred. Thus, the adopted approach can be useful to assess debris flow susceptibility in this and other similar areas. Keywords: Debris Flow, Susceptibility, Tropical mountains; Median bed surface clast size

1. Introduction Debris flow is a very destructive kind of mass movement that may occur in many geomorphological environments (Jakob & Hungr, 2005). There are some definitions of debris

ACCEPTED MANUSCRIPT flow (Costa, 1988; Hungr et al., 2001), but it is generally accepted that they consist of a rapid form of mass movement in which a combination of loose soil, rock, organic matter and water mobilize as a slurry that flows downslope.

T

Until the last decades of the 20th century, debris flow events had not been a matter of much

IP

concern in Brazil. However, recent catastrophic debris flows in mountainous regions of

SC R

southern and southeastern Brazil have called the attention to them due to hundreds of casualties and severe property damages (Kobyiama et al., 2015). Consequently, some studies have started dealing with this subject, many of them related to risk analysis or to the modeling of debris flow susceptible basins (Gramani, 2001; Avellar, 2003; Kanji et al., 2003; Lopes,

NU

2006; Correa et al., 2009; Rocha, 2011). It is important to note that most of those events occurred in mountainous regions covered by the remaining of tropical rainforest stretches

MA

during intense rainfall episodes (e.g.: Kanji et al., 2003).

The headwaters of the upper Velhas river are located on the ridges of Quadrilátero Ferrífero region, the main mineral province in southeastern Brazil. These headwaters present some

D

favorable characteristics for the development of the kind of debris flows that prevails in

TE

Brazil: high declivity; a relatively thick regolith, developed in an Atlantic Tropical Rainforest context; long and narrow drainage channels; high rainfall rates concentrated in the summer

CE P

months. There is no historical record of debris flows in the Upper Velhas basin, but it is important to note that this kind of mass movement usually presents long recurrence time, that can be of thousands of years in Southeastern Brazil (Bierman et al., 2014). As Europeans

AC

settled in this region only by the end of the 17th century, the lack of a historical record can be attributed to an insufficient time span of observation for these poorly active headwaters. Recent debris flow deposits are relatively easy to identify in the field (Pierson, 2005), but older ones are more difficult to be recognized, due to the natural reworking of the thinner components of the debris flow deposits by processes with lower recurrence interval, especially floods. However, some researchers (e.g. Brummer & Montgomery 2003) argued that debris flowcontrolled streams can show distinctive sedimentological and morphometric patterns. Indeed, these authors found in the Rocky Mountains drainage basins (Washington State, USA) a downstream coarsening of bed surface clast size in headwater streams with drainage water lower than 10km2. They interpreted this pattern as a consequence of debris-flow transported processes. According to Brummer & Montgomery (2003), the tendency for downstream

ACCEPTED MANUSCRIPT coarsening in headwater areas where debris flow processes establish the channel gradient can also be correlated to some morphometric attributes, such as channel slope and the unit stream power. This behavior was also found in Alpine streams of Italy (Vianello & D’Agostino,

T

2007).

IP

In this research we intend to verify if the patterns observed in the Rocky Mountains and in

SC R

Alpine basins also occur in a region of different geological, geomorphological, vegetation cover and weather conditions, such as the tropical mountains of the Upper Rio das Velhas basin. Additionally, we incorporate other methods, such as bed surface clast composition and roundness, as well as an investigation of size and composition of large boulders that occur

NU

throughout the study area.

We believe that this integrated methodological approach on the surface sediments in the river

MA

network will help the identification of lag deposits caused by debris flows that can have supposedly occurred at recent geological time in the region. In other regions of Brazil where debris flows have recently been triggered, no historical records of such accidents were made

D

(Kobyiama et al., 2015), not even in the study area. Thus, if our approach proves to be

TE

successful to identify past debris flow events, it will be very important to prevent future accidents here and in other regions with similar characteristics. This kind of study is even

CE P

more relevant in view that many fast growing cities in Brazil have been sprawling towards steeper slopes in mountain flanks.

AC

1.1 The Study Area

The study area is situated in the headwaters of the Upper Velhas river, in Ouro Preto county, Minas Gerais state, southeastern Brazil. These headwaters drain the inner parts of the Mariana Anticline, a mega-fold in the region known as Ferriferous Quadrangle (Alkmim & Marshak, 1998). As this anticline is sculptured by fluvial erosion caused by the Velhas river, the basin is limited in the south and northeast by the Ouro Preto and Antonio Pereira ridges, respectively (Figure 1) that constitute the limbs of the Mariana Anticline. As the resistant rocks that sustain these ridges (quartzites and itabirites) dip outwards with moderate to high angles, they form hogbacks (Huggett, 2011). Thus, the headwaters of the Upper Velhas river drains the other eroded side, the characteristically steep hogback front slope of the Ouro Preto and Antônio Pereira ridges. The tributaries of the Upper Velhas river present long (1 to 5 km) and straight drainage channels in the hogback front slope, that is covered by a relatively thin regolith cover (Costa

ACCEPTED MANUSCRIPT et al., 2014). These characteristics and the high mean annual rainfall in the region (1610 mm.y-1 between 1988 and 2004, with a maximum of 2512 mmy-1 in 1990 – Castro, 2006), also favor the triggering of debris flows. It is worth mentioning that 87% of the rainfall

T

usually is concentrated in the summer months (October to March), and that a very intense

IP

rainfall of 161mmd-1 was recorded in February 1979 (Castro, 2006).

SC R

Archean rocks of Rio das Velhas Supergroup dominate in the Upper Velhas basin, whereas Upper Proterozoic rocks of the Minas Supergroup crop out in a smaller area of the headwaters (Dorr, 1969; Alkmim & Marshak, 1998 – Figure 1).

The Rio das Velhas Supergroup is formed by the Nova Lima and Maquiné groups. The first

NU

one predominates in the basin and is mainly composed of green schists and phyllites and secondarily by quartzites, graywackes, dolomites, talc schists and iron formation (of the

MA

Algoma type). This geological unit constitutes convex hill slopes where Haplic Cambisols develop (IEF, 2005). The Maquiné Group crops out only in the northeastern region, in the hogback front slope of the Antonio Pereira ridge (Figure 1). It is mainly composed of

D

quartzites and quartz-sericite schists (Baltazar et al., 2005) and it sustains a relief of

TE

structurally controlled scarps (IEF, 2005).

CE P

In the region, The Minas Supergroup is composed of two dominant groups (Dorr, 1969; Alkmim & Marshak, 1998): Caraça (quartzites, phyllites, quartz-sericite schists and conglomerates); and Itabira (itabirites, and dolomites). Due to their inherent expressive strength, especially the quartzites and itabirites, they sustain the hogbacks front slopes of

AC

Ouro Preto and Antonio Pereira ridges. Aluminum or iron duricrusts (“canga”) of Paleogene age (Door, 1969) can be found on the itabirites. Most of the Upper Velhas basin is covered by an evergreen forest (Atlantic rainforest) in Nova Lima Group areas. Due to their inherent high weathering resistance, relatively fresh rocks of the Minas Supergroup and Maquiné Group usually crop out. Locally Leptsols can occur (IEF, 2005). Neotropical altitudinal grasslands are the natural vegetation cover for these units (IEF, 2005). The headwaters of the Upper Velhas basin do not show a significant anthropogenic disturbance, since mining or timbering occurs only locally (Freitas, 2010). However, deforestation has increased towards the headwaters as a consequence of fast urban sprawling (Freitas, 2010).

ACCEPTED MANUSCRIPT In the study area colluviums are described, mainly in the hogback front slopes (Costa et al, 2014). They are composed of transported soil, sometimes deposited over a stone line. Costa et al. (2014) dated stone line fragments with the Luminescence Optically Stimulated (LOE)

T

method and found ages between 2880 +/- 465 and 6000 +/- 940 yBP, which implies Mid to

IP

Late Holocene age for the colluvium deposition.

SC R

There is no fluvial terraces in the study area, but Magalhães et al. (2011) recognized three fluvial terraces some kilometers downstream associated with three down-cutting periods: an older (T3) of ~48 kyBP; an intermediate (T2) of ~7.5 kyBP, and a younger (T1) of ~1 kyBP. These authors related T3 and T2 terraces to a Late Pleistocene regional tectonics and the T1

(2005) dated organic sediments with

14

NU

down-cutting to a response to tectonics and climate change in the Holocene. Bacellar et al. C, defining two fluvial river terraces associated with

MA

erosion periods induced by climate change in a drainage basin nearby: the first one yielded ages in the Late Pleistocene-Early Holocene transition (31.34 kyBP and 7.49 kyBP), and the second one a Mid to Late Holocene age (3.67kyBP to 5.3kyBP).

D

Ledru et al. (1996) recognized more humid conditions in Southeastern Brazil between

TE

16kyBP and 11kyBP and between 4kyBP and 3kyBP, which are consistent with the erosion events reported by Bacellar et al. (2005). Horak-Terra et al. (2015) performed a detailed

CE P

palynological and geochronological analysis in the Espinhaço range, north of the study area. They found six phases of climate changing throughout the Holocene, with a dry period between 4.2 kyBP and 1.16 kyBP, and stabilization toward the current climate at 1.16kyBP.

AC

Thus, climatic conditions have varied greatly since the Late Pleistocene in the region, with wetter phase in Early Holocene, and drier phase in Mid to Late Holocene. These changes caused the advance or retreat of the dense vegetation cover, causing periods of erosion and cut terraces (Bacellar et al 2015).

1.2. Theoretical background Debris-flow deposits are easily differentiated from fluvial ones, but as the former usually present higher recurrence period, they can be mobilized and further modified by fluvial processes of lower recurrence period. Debris flow deposits are characteristically massive, not stratified, and poorly sorted (Pettijohn et al., 1973; Pierson, 2005). In contrast to fluvial deposits, they present angular to sub-angular clasts (Pettijohn et al., 1973; Pierson, 2005).

ACCEPTED MANUSCRIPT Several authors have argued that debris flow-dominated channels present distinctive characteristics when compared to fluvial-dominated channels (e.g. Montgomery & Buffington, 2003; Pierson, 2005; Galia & Skarpich, 2015). According to Stock & Dietrich

T

(2003), the transition from debris flow-dominated channels to fluvial-dominated channels is

IP

characterized by an inflection in the relationship between drainage area and channel slope in a log-log graphic. Debris flow-dominated channels tend to be steeper and the declivity

SC R

decreases more slowly with drainage area when compared with fluvial-dominated channels

AC

CE P

TE

D

MA

NU

(Brummer & Montgomery, 2003).

Figure 1: Location and geological map of the Upper Velhas river basin, limited by the Ouro Preto and Antônio Pereira ridges, which conform the limbs of the bleached Mariana anticline. The dotted lines show the limits of the three basins chosen for this study. (a) - Location of Minas Gerais state (MG) in Brazil (in darker gray); (b) Location of the Ferriferous Quadrangle; (c) - Location of the bleached Mariana Anticline in the Ferriferous Quadrangle.

ACCEPTED MANUSCRIPT The unit stream power (), which is the rate of energy expenditure per unit area of the channel, is commonly used in channel hydraulics and river dynamics (Knighton, 1999), since it expresses the stream ability to incise the bed and transport bed load (Golder & Springer,

T

2005). It is directly proportional to discharge and channel slope and inversely proportional to

IP

channel width.

SC R

The unit stream power can be expressed by the following equation (Brummer & Montgomery, 2003): 𝛾.𝑒𝐴𝑑

(1)

𝑊

NU

𝜔= Where:

MA

 = water unit weight (N/m3) A = drainage area (m2)

D

W = bankfull channel width (m)

TE

e and d = empirical values.

This equation can be developed with some simplifying assumptions (Brummer &

CE P

Montgomery, 2003) and it takes the following form: (2)

AC

Where:

𝑒

𝜔 = 𝜌𝑔 (𝑐 ) 𝐴(𝑑−𝑏) 𝑆

S = channel slope

b and c = empirical constants, specific for the basin in question. Constants b and c can be graphically determined by the well-known relationship between channel width (W) and drainage area (A), the former varying as a power function of downstream changes in the latter, according to the equation (Brummer & Montgomery, 2003; Golden, 2005): 𝑊 = 𝑐𝐴𝑏

(3)

Some researchers (e.g. Knighton, 1999; Brummer & Montgomery, 2003) found out that the unit stream power (𝜔) increases downstream in headwater channels and shifts to a decreasing trend, when lower-gradient alluvial areas are reached.

ACCEPTED MANUSCRIPT Differently to what has been predicted in the literature for rivers, the median bed surface clast size (D50) of headwater channels sometimes presents a coarsening downstream pattern (Solari & Parker, 2000). Indeed, Brummer & Montgomery (2003) studied four drainage basins in the

T

Rocky Mountains, Western Washington (USA), and found that D50 increases until the

IP

drainage area reaches 10 km² and then decreases as the drainage area increases. The empirical equation that expresses the relationships between D50 (in m) and drainage area 𝐷50 = 𝑓𝐴 𝑔 Where:

(4)

NU

f, g = empirical coefficients

SC R

<10 km2 assumes the form (Brummer & Montgomery, 2003):

Although there is no consensus about the coarsening downstream pattern (Golder & Springer, 2006), similar D50 and drainage area relations were also found in other regions, even for other

MA

clast diameters (Table 1).

power function (equation 4). g 0.30 0.297 0.213 0.329 0.321 0.328 0.434

TE

f 0.074 0.087 0.032 0.176 0.212 0.378 0.010

CE P

Di D50 D50 D10 D84 D90 Dmax D50

D

Table 1: Empirical relationships between diameter (Di) and drainage area (A) found by linear regression by r2 0.71 0.64 0.63 0.60 0.57 0.58 0.39

Reference Brummer & Montgomery (2003) Vianello & Agostino (2007) Vianello & Agostino (2007) Vianello & Agostino (2007) Vianello & Agostino (2007) Vianello & Agostino (2007) Leigh (2010)

AC

Vianello & Agostino (2007) interpreted the downstream coarsening pattern in Alpine headwater streams as a consequence of washing of small diameters clasts. They pointed out that this inverse selective transport mechanism would occur where cobbles and boulders were available in the channel substratum. On the other hand, Leigh (2010) interpreted the downstream coarsening of D50 of small streams in the Southern Blue Ridge Mountains of Western North Carolina as a result of the prevalence of colluvial additions of fine sediment from hillslopes small streams. Brummer & Montgomery (2003) interpreted the shift from coarsening to fining as a result of a change in transport capacity, that is initially greater than the supply of sediments and changes to supply of sediments greater than this capacity. In other words, the shift occurs when the condition ‘bed surface sediments carried by debris-flows’ changes ‘to sediments transported by fluvial processes’. Thus, the analysis of D50 variation along channels can be a

ACCEPTED MANUSCRIPT good hydraulic indicator of debris flow processes, together with other hydraulic parameters, such as channel slope and unit stream power (Brummer & Montgomery, 2003). 2. Material and Methods

IP

T

2.1. Fluvial channel data

Initially, field trip and satellite image data were used to find adequate drainage channels in

SC R

order to study their susceptibility to debris-flows. As discussed previously, the Upper Velhas river basin was chosen, since the conditions are favorable to this type of mass movement. Several tributaries of this river were analyzed and some channels were selected on the basis of

NU

some obligatory conditions, such as: good accessibility; similar geological and geomorphological characteristics; small disturbance by human activities, such as mining, timbering and agriculture; and broad range of drainage areas, at least up to 30 km2. A more

MA

preferable condition would be the main stream channel itself, avoiding sampling of tributaries, but this proved to be impossible. A segment of Upper Velhas river channel and

D

two tributaries were chosen for this study.

TE

Following Brummer & Montgomery (2003) recommendations, sampling was carried out in fluvial segments with channel length/width ratio between 10 to 20, in order to avoid

CE P

disturbances caused by plant debris, tributary junctions, mining or timbering. The selected channel segments should not have pools or falls that could disturb or make the sediment transport difficult.

AC

Drainage areas (A) were determined from topographic maps (1:25,000 scale) in a GIS environment. Length (L), width (W), depth (D), channel average slope (S) and bed surface clast size (D50) of each channel segment were obtained following Brummer & Montgomery (2003) recommendations. We adopted a minimum distance of 100 m between two consecutive sampling points in order to obtain enough data in a reasonable scale. As Brummer & Montgomery (2003) did not find any distinctive pattern for median subsurface clast size, we decided to analyze only bed surface clasts, which yielded better results for these authors. The bed clast sampling followed Wolman’s (1954) pebble count method, which is the analysis of the relative area covered by coarse particles of the river bed. The method consists of sampling 100 clasts with medial axis larger than 2 mm, in a grid established by pacing transversely the entire width of the drainage channels. The grid size (maximum 1x1m) varied according to the channel width. Following Wolman’s (1954)

ACCEPTED MANUSCRIPT recommendations, each pebble beneath the tip of the toe is collected without looking at it, so as to guarantee a random sampling. The medial axis of each clast is measured in the field. The clasts were classified according to

T

lithology and roundness, the latter following the classification of Pettijohn et al. (1973).

IP

Following Wolman’s (1954) recommendations, clast medial diameter accumulated frequency

SC R

curves allowed the determination of D16, D50 and D84 for each sample.

Channel slope and bank full width at each sampling point were measured with tape and inclinometer.

NU

2.2. Boulder Distribution Analysis

Boulders of large dimensions and with diverse lithological composition were investigated in

MA

the field. Many of these boulders are allochthonous and present sizes that are incompatible with the bed-load transport competence of the creeks. Thus, boulders were lithologically described and their medial clast diameter was measured in the field. Only boulders with

D

average diameter larger than 1 m were considered in this analysis.

TE

Sediments under six selected boulders were collected for dating with Luminescence Optically Stimulated (LOE) technique in order to establish the probable age of deposition of the

CE P

boulders, since this technique reveals the elapsed age since the sediments were exposed to sunlight. Dating was done following the SAR protocol (Wallinga et al., 2000), with ten aliquots for each sample. The samples were collected with 35 cm-long PVC tubes at 5-cm

AC

depth, where the sediments do not show evidences of remobilization (cut and fill structures, for example). 3. Results

It is very difficult to access the drainage basin areas of the UpperVelhas river, due to the steep relief, with deeply incised drainage system, in the most part covered by a dense evergreen forest. Similar to what occurred to Brummer & Montgomery (2003), it was impossible to select only one stream channel to study, which would be the ideal situation. Indeed, it was only possible to select two main channels in the drainage basins that are more easily accessed: third-order São Bartolomeu and Cardoso creek drainage basins (Figure 2), according to Strahler’s (1952) classification. The drainage areas are small (< 6 km2) and not enough for this type of study, since the transition from coarsening to fining downslope of surface clasts in previous studies occurs with drainage areas of at least 7 km2 (Vianello & D’Agostino, 2007) or 10 km2 (Brummer & Montgomery, 2003). Thus, a larger basin, of the fifth order, was

ACCEPTED MANUSCRIPT included in this study, a segment of the Upper Velhas river, with drainage areas between 10 and 40 km2, towards which the São Bartolomeu and Cardoso creeks drain (Figure 2). It was impossible to study only the Upper Velhas river because its headwaters with drainage areas

T

up to 10 km2 present some degree of human disturbance. It is worth mentioning that the three

IP

selected basins present similar geological and geomorphological characteristics and

SC R

insignificant degree of anthropogenic disturbance.

Aligned hills built on Nova Lima Group schists prevail in these basins, while mountainous crests on more resistant units (quartzites, itabirites and cangas) crop out only above their headwaters (Figure 1). These harder units support a hogback formed by the Mariana anticline

NU

limbs, locally named Ouro Preto and Antonio Pereira ridges, respectively south and northeast of the region (Figure 1).

MA

In the three basins, 87 sampling points (Figure 2) were selected in appropriate channel reaches of the respective main stream in drainage areas of 0.05-5.5 km2 (São Bartolomeu and

AC

CE P

TE

were collected and analyzed.

D

Cardoso creeks) and of 10-40 km2 (Upper Velhas river). In each reach, bed surface clasts

Figure 2: Sampling points in reaches of the stream channel of three selected drainage basins.

The calculated D50 values for 100 clasts of each sampling point were plotted against the respective drainage area for the three drainage channels (Figure 3a). A direct relationship can

ACCEPTED MANUSCRIPT be observed for areas between 0.05 and 5.5 km2, which shifts to an inverse relationship for areas larger than 10 km² (Figure 3a). As it was impossible to collect samples in areas between 6 and 10 km2, it is not possible to establish the exact point of inflexion. Likewise, a similar

T

relationship was found for the D16 and D84 (Figure 3b e 3c), especially D84, with a higher r2.

IP

Two domains are defined when plotting channel slope versus drainage area (Figure 3d). Local slope tends to be steeper for drainage areas of circa 1 km2 and gentler for drainage areas larger

SC R

than 10 km2, with a transition zone with scattered data between 2-5 km2 (Figure 3d). There is also a direct relationship between channel width and drainage area for the São Bartolomeu and Cardoso creeks, but for the Upper Velhas river it is not so evident (Figure 3e). Even so,

NU

the correlation analysis allowed the determination of parameters b and c (Equation 2), used to

AC

CE P

TE

D

MA

determine the unit stream power.

Figure 3: Plots showing the relationship between drainage area and some hydraulic parameters for the three drainage basins: Black circles = Rio das Velhas basin; Gray circles = Cardoso basin; White circles = São Bartolomeu basin. a) D50 versus drainage area. The downstream coarsening regression of surface D50 for basins <5.5 km2 follows the relation D50 = 0.0255A0.5741 (r2 = 0.70). b) D16 versus drainage area. The downstream coarsening regression of surface D16 for basins <5.5 km2 follows the relation D16 = 0.0119A0.4366 (r2 = 0.58). c) D84 versus drainage area. The downstream coarsening regression of surface D84 for basins <5.5 km2 follows the relation D84 = 0.0529A0.5298 (r2 = 0.78). d) Channel slope versus drainage area; e) Channel width versus drainage area. The regression of W for the three basins are: Rio das Velhas - w =3.306A 0.19 (r2=0.06); Cardoso w=1.950A 0.373 (r2=0.86); São Bartolomeu - w=1.286A 0.383 (r2=0.74) f) Stream power versus drainage area.

ACCEPTED MANUSCRIPT The relationship between the unit stream power and drainage area is not well defined, but a direct relationship apparently occurs between drainage areas of 0.6 and 4.0 km2 (Figure 3f). Above 10 km2 a shift in this behavior is observed, although this transition is not evident, due

T

to the lack of data for areas between 6 and 10 km2.

IP

The bed surface clast samples (100 clasts for each of the 87 selected reaches) were classified

SC R

in the field according to their composition and roundness. The data show that (Nova Lima Group) schists are the dominant lithology. Clasts from the Moeda Formation quartzites, Cauê Formation itabirites (metamorphosed banded iron formation), and quartz veins occur in minor

AC

CE P

TE

D

MA

NU

proportions (Figure 4). Clasts composed of iron duricrusts (“canga”) are the least frequent.

Figure 4: Plots of bed load clast compositions (graphs on the left) and roundness (right) versus drainage area. (a) São Bartolomeu creek, (b) Cardoso creek, and (c) Upper Velhas river.

ACCEPTED MANUSCRIPT

Most of the drainage channel surface clasts are angular or sub-angular, according to Pettijohn et al. (1973) roundness classes (Figure 4), which is expected for small drainage basins, where

T

the distance of transport is small.

IP

All the boulders with medial diameter greater than 1 m that crop out on slopes (colluvium or

SC R

talus) or valley bottoms (channel floor, floodplain and fluvial terraces) were catalogued and mapped (Figure 5). Although it is impossible to record most of the boulders of the three basins, due to lack of roads and also to difficulty of access, we believe that the 231 described boulders are somewhat representative of their distribution. Quartzite is clearly the dominant

NU

boulder composition (Figure 6) and, to a lesser extent, quartz vein and schist (Figure 5). The boulder sizes are variable, but it is not uncommon boulders of 6 to 7 m in diameter.

MA

LOE dating was performed in order to determine the depositional ages of the transported boulders, i.e., of the ones with proven allochthonous lithologies. The sediment samples for

D

dating were collected under the boulders along the São Bartolomeu creek (Figure 5), because

TE

for this basin previous LOE dating of stone lines under colluvium in the hogback front slope are available (Costa et al., 2014). Six larger boulders were selected, since in this situation the

CE P

remobilization of sediments after deposition is more difficult. Dating reveals a wide range of depositional ages, between 0.55 and 30kyBP (Table 2 and Figure 5).

AC

Table 2: LOE dating of sediments collected under boulders. Sample 1 2 3 4 5 6

Annual dose (mGy/y) 2,050 +/230 2,610 +/300 3,150 +/320 2,820 +/360 1,910 +/220 2,160 +/285

Accumulated dose (Gy) 61.3 46.6 11.2 1.5 4.1 11.2

Age (y) 30,000 +/- 4,800 17,900 +/- 2,900 3,550 +/540 550 +/95 2,150 +/360 5,170 +/940

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 5: Location map of the boulders described in the study area. Note the great number of allochthonous

AC

CE P

quartzite boulders not only to the south, in the flanks of the Ouro Preto ridge, but also along the channels.

Figure 6: Quartzite boulders that crop out along the channel floor or valley bottoms.

ACCEPTED MANUSCRIPT 4. Discussions The spatial distribution of bed surface clast size (D50) shows coarsening downstream in drainage areas of 5.5 km2 and fining downstream for drainage areas larger than 10 km2

T

(Figure 3a). A similar pattern was found with D16 and D84 (figures 3b and 3c), especially the The coarsening downstream is not a usual

SC R

allochthonous) to bedload transport events.

IP

latter (higher r2), which can be attributed to a lower mobility of greater clasts (many of them

sedimentological behavior, but it has been described in headwater channels of other mountainous regions, known to be susceptible to debris flow movements (Brummer & Montgomery, 2003; Vianello & D’Agostino, 2007). In order to confirm this trend, the data

NU

obtained from the graph median bed surface clast size (D50) versus drainage area was plotted on the graph proposed by Brummer & Montgomery (2003) (Figure 7). According to these

MA

authors, the transition between debris flow to fluvial-dominated channels occurs for drainage areas of 1 to 10 km2. The lack of data for drainage areas between 6 and 10 km2 makes the precise determination of where this transition occurs difficult. The sampling points spread

D

below the curve proposed by Brummer & Montgomery (2003), but this can be caused by

AC

CE P

TE

certain characteristics of the study area, e.g., it is located in a tropical evergreen forest.

Figure 7: Relationship between D50 and drainage area for the basins of this study (SB = São Bartolomeu; CA = Cardoso; RV= Upper Rio das Velhas) was plotted on the graph proposed by Brummer & Montgomery (2003). The vertical arrowed line represents the natural variability found by these authors.

ACCEPTED MANUSCRIPT Although there is not a perfect coincidence with Brummer & Montgomery (2003) model, the general behavior is similar, so it is possible to interpret by analogy that surface clasts originated from debris flows in drainage areas up to 6 km2 and from fluvial deposits in

T

drainage areas larger than 10 km2.

IP

Channel slope is steeper for small drainage areas and apparently shows an inverse relationship

channel slope is significantly lower (Figure 3d).

SC R

for drainage areas of 2-3 km2 (Figure 3d). Above this interval this pattern is not clear, but the

The unit stream power apparently presents a direct relationship with drainage areas between 0.6 and 4 km2 (Figure 3f). For drainage areas greater than 10 km2, there is a shift in this

NU

behavior, although this transition is not well defined for the study area. Anyway, the trends observed for drainage areas versus channel slope or versus unit stream power do not rule out

MA

the trend found by Brummer & Montgomery (2003) in four Rocky Mountains drainage basins. Accordingly to their interpretation, steeper slopes, with higher energy flows facilitate the transport of sediments by debris flows. This pattern changes downstream, as the flows are

TE

D

less energetic and the slopes gentler, with a predominance of fluvial processes. As previously commented, geological units with distinctive characteristics (itabirites,

CE P

quartzites and iron duricrusts) crop out exclusively in the surroundings of the crests of the Ouro Preto and Antônio Pereira ridges, above the headwaters of the basins of this study. This means that the clasts of these units could be good tracers of transport distance. Therefore, we expect that clasts with these compositions would prevail in channels near the ridges, i.e.,

AC

south of the Upper Velhas river and Cardoso and São Bartolomeu creeks, and in the northeastern part of the Upper Velhas river, where it drains the front slopes of the Ouro Preto and São Bartolomeu ridges respectively (Figures 1 and 2). The data show that Nova Lima Group schist clasts predominate as bed surface clasts along the three drainage channels (Figure 4). This pattern was expected, since these rocks outcrop in most of the region (Figure 1). However, there are considerable amounts of quartzite and itabirite clasts throughout these channels, suggesting a significant transport for several kilometers far from the source area (Figure 4). When the bed surface clasts are grouped according to their susceptibility to weathering, mainly by abrasion, it is possible to note that more resistant ones (quartzite + quartz vein) are less abundant in drainage areas smaller than 2 km2 and are more abundant in drainage areas greater than 10 km2. On the other hand, less

ACCEPTED MANUSCRIPT resistant (schist) clasts show an inverse relationship (Figure 8a). It is important to note that for drainage areas between 2 and 6 km2, the data show great dispersion. Surface clasts are classified as angular and sub-angular according to Pettijohn et al. (1973)

T

(Figure 4). Increase in roundness can be seen downstream (Figure 4). When we group very

IP

angular and angular clasts and rounded and well rounded clasts, it is possible to note a similar trend, with smaller roundness degree for drainage areas smaller than 2 km2, and larger

SC R

roundness degree for drainage areas larger than 10 km2, with a significant dispersion between these values (Figure 8b), suggesting that a different transport process can be acting in larger drainage areas. As debris flow deposits tend to be less weathered and to contain more angular

NU

clasts when compared to fluvial ones, this pattern shift could be another evidence of ancient debris flow deposits in the headwaters of these basins.

MA

The spatial distribution of large and angular boulders that are randomly oriented throughout the basins (Figure 5) permits the interpretation of their provenance, since they were found in

TE

D

two geomorphological contexts:

(a) on steep slopes, as talus or colluvium deposits which occur on the hogback front slopes of

CE P

the Ouro Preto and Antônio Pereira ridges (in the limbs of the bleached Mariana anticline). These are mainly formed by the fall or rolling of more resistant rocks that support the hogback crests of these ridges (especially the quartzites of the Moeda Formation);

AC

(b) on valley bottoms (channel floor, floodplains and fluvial terraces) mainly on areas of occurrence of the Nova Lima Group. They are composed mainly of quartzites and quartz veins (Figure 5), the source area of the former is up to 6 km upstream. The huge size of these boulders, some of them reaching 8 m of medial diameter, is incompatible with the competence of the drainage system. Thus, their occurrence is another evidence of debris-flow events. These debris-flow events of larger recurrent periods could have been partially mobilized by smaller energy fluvial events of lower recurrence period, but these processes do not have enough energy to hide the debris flow lag deposits.

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

Figure 8: Bed surface clast composition and roundness degree versus drainage area.

LOE dating of sediments collected under the boulders distributed along the São Bartolomeu channel yielded a wide range of ages, between 30 kyBP and 0.55 kyBP, with two ages in the Upper Pleistocene and four in the Mid to Late Holocene (Table 2). Previous studies involving 14

C and LOE dating of fluvial terraces and colluvium show that at least two or three incision

phases related to erosion events occurred in the study area, two during the Late Pleistocene and one during the Mid to Late Holocene (Bacellar et al., 2005; Costa et al., 2014) or Late Holocene (Magalhães et al., 2011). Therefore, it is possible that debris-flow events occurred in the study area as a consequence of climate change during these epochs, causing advances and retreats in the vegetation cover and favoring erosion and mass movements. Deforestation that is currently occurring in Brazilian municipalities as a consequence of fast urban spreading towards mountainous areas can have the same effect and can expose these

ACCEPTED MANUSCRIPT mountainous regions to severe geological risk, especially when associated with extreme rainfall.

IP

T

5. Conclusions

This paper brought a new approach to the assessment of debris flow susceptibility, integrating

SC R

the analysis of bed load clast and boulder characteristics and spatial distribution and channel morphometry. This approach allowed the analysis of the susceptibility to possible debris flows in areas lacking historical records of this kind of mass movement.

NU

Sedimentological data collected from three mountainous drainage channels showed a coarsening downstream pattern in drainage areas up to 6 km2 and a fining downstream pattern

MA

in drainage areas of more than 10 km2. Although rarely described, this sedimentological pattern has been reported in other mountainous regions of non-tropical areas, such as the Rocky Mountains (Brummer & Montgomery, 2003) and the Italian Alps (Vianello &

D

D’Agostino, 2007). The shift of coarsening to fining has been attributed to the change from

TE

debris flow-dominated channels to fluvial-dominated channels (Brummer & Montgomery, 2003).

CE P

Morphometric parameters obtained in this study suggest that anomalies that occur in drainage areas between 3 and 4 km2 may be interpreted as the transition of these two domains, which occurs simultaneously with the reduction of channel slope and, perhaps, the specific stream

AC

power, as described by Brummer & Montgomery (2005). The analysis of allochthonous bed load clasts found along the drainage channels provided important information about source area and transport. It is usually expected for fluvial-dominated channels that the concentration of weathering resistant clasts and their roundness degree should increase gradually downstream. However, we observed that this transition is more abrupt and occurs in drainage areas of 2 to 5 km2, which may be another evidence of change in transport mechanisms, from debris flow to fluvial. Other important factor to be highlighted is the randomly oriented, angular boulders of large diameter found along drainage channels and valley bottoms. Huge quartzite boulders are found in small channels up to 6 kilometers far away from the source area. As the creeks may not have the capacity to carry such boulders, they may have been transported by debris flow events. LOE dating suggests that these boulders were transported in Upper Pleistocene to Holocene events, supporting other studies that point out some erosion events with similar ages

ACCEPTED MANUSCRIPT associated with climate changes and vegetation advance and retreat (Bacellar et al., 2005; Costa et al., 2014; Magalhães et al., 2011). Nowadays, deforestation can have the same effect, exposing these regions to debris flow risk.

T

In short, we propose a new and simple approach to assess the probable occurrence of debris

IP

flows in recent geological time in mountainous regions. Field surveys and simple laboratory

SC R

tests can help the identification of the debris flow susceptibility in headwater channels in mountainous regions. It is worth mentioning that this method still needs to be better tested in other areas. The development of procedures to assess debris flow susceptibility is important in countries such as Brazil, where debris flows have recently occurred in areas with no historical

NU

record of such movements.

MA

Acknowledgments

We wish to thank UFOP, FAPEMIG, CNPq and CAPES for the financial support to this

TE

D

work.

CE P

REFERENCES

Alkmim F.F., Marshak S., 1998. Transamazonian orogeny in the Southern São Francisco Craton region, Minas Gerais, Brazil: evidence for Paleoproterozoic collision and collapse in the Quadrilátero Ferrífero. Precambrian Research, 90: 29-58.

AC

Avelar, A. S., 2003. Iniciação e desenvolvimento de fluxos detríticos em encosta: ensaios triaxiais de choque e em caixa experimental abordando o caso do Soberbo (RJ). (DSc Thesis) Universidade Federal do Rio de Janeiro.

Bacellar LAP. 2000. Condicionantes geológicos, geomorfológicos e geotécnicos dos mecanismos de voçorocamento na bacia do rio Maracujá, Ouro Preto, M.G., (DSc Thesis) Universidade Federal do Rio de Janeiro. Bacellar, L.A.P.; Coelho Netto, A.L., Lacerda, W.A., 2005. Controlling factors of gulling in the Maracujá

catchment,

Southeastern

Brazil.

Earth

Surf.

Process.

Landforms,

30:1369-1385.

doi: 10.1002/esp.1193 Bagnold, R. A., 1966. An approach to the sediment transport problem from general physics. Geological Survey Professional Paper, Washington, 42p

ACCEPTED MANUSCRIPT Baltazar, O. F., Zucchetti, M., 2005. Lithofacies associations and structural evolution of the Archean Rio das Velhas greenstone belt, Quadrilátero Ferrífero, Brazil: A review of the setting of gold deposits. Ore Geology Reviews. 32(3/4), 471-499. doi:10.1016/j.oregeorev.2005.03.021

T

Bierman, P., Fernandes, N., Sosa-Gonzalez, V., Rood, D. H., Massey, C., 2014. Cosmogenic 10-Be

IP

analysis of debris flow boulders and sediment from Brazil indicates long return times, Geological Society of America Abstracts with Programs, abstract no: 247773

SC R

Brummer, C. J., Montgomery, D. R., 2003. Downstream coarsening in headwater channels. Water Resources Research, 39 (10), 1-14. doi: 10.1029/2003WR001981

Castro, J.M.G., 2006. Pluviosidade e movimentos de massa nas encostas de Ouro Preto. (MSc

NU

Dissertation). Universidade Federal de Ouro Preto.

Correa, G. P., Kobiyama, M., Giglio, J. N., Michel, G. P., Tasca, F. A., Goer, R. F., 2009. Análise

MA

geométrica para estimativa de volume em fluxo de detritos, in: V Seminário Internacional de Defesa Civil, São Paulo.

Costa, F.O., Bacellar, L.A. P., Castro, S.S., Resende, F.R.M., Silva, S.F., 2014. Morfopedogênese de

D

zona escarpada na alta bacia rio das Velhas, Quadrilátero Ferrífero, MG: Análise integrada de dados

TE

físicos, químicos e mineralógicos. Revista de Geociências 33, 403-415. Costa, J. E., 1988. Rheologic, morphologic, and sedimentologic differentiation of water floods,

CE P

hyperconcentrated flows, and debris flows, in Baker, V.E., Kochel, C.R., and Patton, P.C., editors, Flood geomorphology. John Wiley and Sons, New York, pp. 113-122. Dorr, J. v. N., 1969. Physiographic, stratigraphic and structural development of Quadrilátero Ferrífero,

AC

Minas Gerais, Brazil. USGS. Prof. Paper, 641-A, Washington. 110 p. Freitas, S.M.A.C., 2010. Quantificação da Infiltração e da Recarga de Aqüíferos em Microbacias do Alto Rio Das Velhas M.G. (MSc Dissertation). Universidade Federal de Ouro Preto. Galia, T., Skarpich, V. 2015. Do the coarsest bed fractions and stream power record contemporary trends in steep headwater channels? Geomorphology. doi: 10.1016.j.geomorph.2015.07.047 Golder, L. A., Springer, G.S., 2006. Channel geometry, median grain size, and stream power in small mountains streams. Geomorphology 78: 64-76. doi:10.1016/j.geomorph.2006.01.031 Golden, L.A. 2005. Lithologic controls of headwater stream morphology in the eastern Appalachian Plateau, Western Virginia. MSc Thesis. Ohio University. 118p. Gramani, M. F., 2001. Caracterização geológica-geotécnica das corridas de detritos (“Debris Flows”) no Brasil e comparação com alguns casos internacionais. (MSc Dissertation). Universidade de São Paulo. São Paulo..

ACCEPTED MANUSCRIPT Horák-Terra, I., Cortizas, A.M., Luz, C.F.P., Silva, A.C., Vidal-Torrado, P., 2015. Holocene climate change in central–eastern Brazil reconstructed using pollen and geochemical records of Pau de Fruta mire (Serra do Espinhaço Meridional, Minas Gerais). Palaeogeography, Palaeoclimatology, Palaeoecology.

T

437, 117-131. doi:10.1016/j.palaeo.2015.07.027

SC R

Hungr, O., Evans, S.G., Bovis, M., and Hutchinson, as euro

IP

Huggett, J.R., 2011. Fundamentals of Geomorphology, 3rd ed., Routledge, New York.

J.N., 2001, Review of the classification of landslides of the flow type. Environmental and Engineering Geoscience, 7,. 221-238.

IEF 2005. Zoneamento ecológico da área de proteção ambiental estadual Cachoeira das Andorinhas,

NU

Ouro Preto, Minas Gerais. Instituto Estadual de Florestas. Relatório do meio físico. Universidade Federal de Viçosa. 94p

MA

Jakob, M., Hungr, O, 2005. Debris-flow Hazards and Related Phenomena. Springer Praxis Books. Chichester.

Kanji, M.A., Massad,F., Cruz, P.T., 2003. Debris flows in areas of residual soils: occurrence and

TE

Earthfills, Sorrento, 2, 1-11.

D

characteristics. International Workshop on Occurrence and Mechanism of Flows in Natural Slopes and

CE P

Knighton, A.D., 1999. Downstream variation in stream power. Geomorphology 29, 293–306. Kobiyama, M.; Michel, G.P.; Engster, E.C.; Paixão, M.A., 2015. Historical analyses of debris clow disaster occurrences and of their scientific investigation in Brazil. Labor & Engenho, Campinas. Brasil, 9 (4), 76-89.

AC

Ledru, M.P., Braga, P.I.S, Soubiès, F., Fournier, M., Martin, L., Suguio, K., Turcq, B., 1996. The last 50.000 years in the Neotropics (Southern Brazil): evolution of vegetation and climate. Palaeogeography, Palaeoclimatology, Palaeoecology. 239-257. Leigh, D. S., 2010. Morphology and Channel Evolution of Small Streams in the Southern Blue Ridge Mountains of Western North Carolina Southeastern Geographer. V. 50, n 4, pp. 397-421. doi: 10.1353/sgo.2010.0010 Lopes, E. S. S., 2006. Modelagem Espacial Dinâmica Aplicada ao Estudo de Movimentos de Massa em uma Região da Serra do Mar Paulista, na escala de 1:10.000. (PhD Thesis). Universidade Estadual Paulista. Magalhães Junior, A.P, Cherem, L. F. S., Barros, L. F. P., Santos, G. B., 2011. OSL dating of sediments from a mountainous river in southeastern Brazil: Late Cenozoic tectonic and climatic implications. Geomorphology. 132 (3-4), 187-194. Pettijohn, F. J., Potter, P. E., Siecer, R., 1973. Sand and sandstone., Springer-Verlag. New York.

ACCEPTED MANUSCRIPT Pierson, T. C., 2005. Distinguishing between Debris Flows and Floods from Field Evidence in Small Watersheds. USGS Fact Sheet 2004-3142. U.S. Geological Survey. Rocha, H. L., 2011. Aplicação do modelo Flo-2d para simulação de fluxos de detritos na bacia do rio

T

Cunha, rio dos Cedros/SC. (PhD Thesis). Universidade Federal de Santa Catarina.

IP

Solari, L., Parker, G., 2000. The Curious Case of Mobility Reversal in Sediment Mixtures. J. Hydraul.

SC R

Eng.,126 (3), 185-197.

Stock, J., Dietrich, W.E., 2003. Valley incision by debris flows: Evidence of a topographic signature. Water Resources Research. 39 (4), 1089. doi:10.1029/2001WR001057

Strahler, A.N., 1952. Hypsometric (area-altitude) analysis and erosional topography. Geological

NU

Society of America Bulletin, 63, 1117-1142.

Vianello, A.; D´Agostino, V., 2007. Bankfull width and morphological units in an alpine stream of the

MA

dolomites (Northern Italy). Geomorphology. v.1. 83 (3–4):30. doi:10.1016/j.geomorph.2006.02.023 Wallinga, J., Murray, A., Wintle, A., 2000. The single-aliquot regenerative-dose (SAR) protocol

D

applied to coarse-grained feldspar. Radiation Measurements 32, 529–533. Wolman, M. G., 1954. A Method of Sampling Coarse River-Bed Material.Transactions, American

AC

CE P

TE

Geophysical Union. 3 (6), 951-956.