tect20238-sup-0002-supplementary

advertisement
1
Supporting information for
2
3
Investigating multiple fault rupture at the Salar del Carmen segment of the Atacama Fault System
(northern Chile): Fault scarp morphology and knickpoint analysis
4
5
Authors
6
1.
Oktawian Ewiak (corresponding author)
7
Department of Geodynamics
8
GeoForschungsZentrum Potsdam
9
Telegrafenberg
10
D-14473 Potsdam
11
oewiak@gfz-potsdam.de
12
13
2.
Pia Victor
14
Department of Geodynamics
15
GeoForschungsZentrum Potsdam
16
Telegrafenberg
17
D-14473 Potsdam
18
pvictor@gfz-potsdam.de
19
20
3.
Onno Oncken
21
Department of Geodynamics
22
GeoForschungsZentrum Potsdam
23
Telegrafenberg
24
D-14473 Potsdam
25
oncken@gfz-potsdam.de
26
Journal:
27
Tectonics (2015)
1
28
Supporting information outline
29
1. Introduction
30
2. Sensitivity tests
31
3. Calculation of synthetic SRL
32
4. File descriptors and supplementary figure captions
33
34
35
36
37
38
39
40
41
42
43
44
45
46
2
47
1. Introduction
48
The supplementary material to this article contains two short text passages describing a) the
49
sensitivity tests carried out to examine the effects of varying parameters on the routine for
50
calculation of gradient along profile, and to estimate an age threshold up to which our approach
51
reliably detects breaks in slope in the analyzed profiles, and b) the calculation of synthetic rupture
52
lengths from previously derived paleomagnitudes. These descriptions are provided in the
53
paragraphs below.
54
55
2. Sensitivity tests
56
To examine the effects of varying parameters on the routine for calculation of gradient along
57
profile, and to estimate an age threshold up to which our approach reliably detects breaks in slope
58
in the analyzed profiles, we conducted a series of tests. For this purpose, we created synthetic
59
profiles with a length of 100 m. We varied the initial scarp height, the background slope angle α
60
and the initial maximum scarp slope angle θ using the Matlab code Scarp Diffusion Lab (SDL)
61
[Hilley and Arrowsmith, 2003]. Subsequently, we modeled the degradation at different time steps
62
using the forward modeling function in SDL. Modeling of scarp degradation is strongly
63
dependent on the diffusion constant κ. We have chosen two end member diffusion constants
64
(κ1 = 0.46 m²/ka, κ2 = 0.046 m²/ka) proposed as reasonable values for the hyperarid climate
65
environment of the N-Chilean Atacama desert by González and Carrizo [2003]. After forward
66
modeling of scarp degradation we exported the synthetic data and analyzed it using our sliding
67
window approach.
68
We have tested two initial maximum scarp slopes θ1 = 80° and θ2 = 60° with a background slope
69
α = 0° and initial scarp heights of 1 m and 3 m. Degradation was modeled for time steps of 500 a,
3
70
1 ka, 10 ka, 20 ka, 40 ka, 60 ka, 80 ka, and 100 ka using diffusion constants κ1 and κ2.
71
Subsequently, we calculated gradient along the synthetic profiles using window sizes between 3
72
and 25. The initial scarp height had no effect on the obtained results. Variation of window size
73
also had no significant impact on the detection of slope breaks, but still proved useful to decrease
74
noise in real data. Troughs in gradient data were recognizable over the whole range of modeled
75
ages (Figures S4 – S5), but profiles with model ages over 40 ka (κ1) only introduced minor
76
changes to the output as erosion acts strongest on steep surfaces. Using the lower diffusion
77
constant κ2 (κ2 = κ1 * 0.1) resulted in steeper troughs for higher ages (Figures S6 – S7). This
78
implies that erosion has a significant impact on our approach for detection of slope breaks. The
79
initial scarp slope angle θ influences the width of the troughs in gradient data. For very low initial
80
dip angles, e.g., 10°, the resulting gradient data would show broad valleys instead of steep
81
troughs due to the stability of gradient over distance. However, for realistic initial values
82
(θ1 = 80°, θ2 = 60°) gradient data shows prominent troughs (Figures S4 – S7).
83
In a subsequent test, we used profiles with an initial maximum scarp slope of 60°, but changed
84
the background slope α to 4°. All other parameters were kept the same. The gradient data shows
85
prominent troughs up to 60 ka for degradation based on constant κ1 for all window sizes
86
(Figure S8). Using κ2, gradient data displays prominent troughs for the entire range of tested ages
87
up to 100 ka (Figure S9). Thus, for an normal fault with a realistic dip angle of ~60°, our sliding
88
window approach is capable of reliably detecting slope breaks in fault scarp data over the entire
89
range of time steps.
90
Variation of window sizes was mainly used to improve the signal to noise ratio. Smoothing
91
increased with window size. We note that the output slope angle was observed to decrease with
92
increasing window size as the differentiation is carried out for an increasing amount of data
4
93
points. The most accurate measurement of slope angle is accomplished by using the smallest
94
window size. However, to derive the number of events recorded in profile data, we focused on
95
detecting breaks in gradient rather than measuring absolute slope angles.
96
97
3. Calculation of synthetic SRL
98
Synthetic SRL were calculated from paleomagnitudes (Figure S10; Table 1, equation (6)), which
99
were previously derived from (a) the LDI at the fault scarp and (b) displacement at knickpoints
100
(KnpD) to compare them with the mapped SRL in the study areas. SRL based on LDI range from
101
19.0 km to 34.2 km in the CJS, and from 12.9 km to 28.1 km in the LNS. SRL based on KnpD
102
range from 8.7 km to 23.8 km for the CJS, and from 7.7 km to 19.8 km for the LNS. The
103
synthetic SRL based on paleomagnitudes calculated from LDI and KnpD are in reasonable
104
agreement with the mapped total length of the CJS (19 km). In case of the LNS, all synthetic SRL
105
are larger than the mapped segment. Thus, for large earthquakes with Mw ~ 7, we assume an
106
involvement of adjacent segments.
107
108
109
110
111
112
5
113
4. File descriptors and supplementary figure captions
114
Filename: tect20238-sup-0003-pS01.tif
115
File descriptor: Aerial view of CJS
116
Figure S1: Aerial view of the area depicted in Figure 2a. Aerial photographs were acquired with
117
the Modular Airborne Camera System, developed by the DLR (German Aerospace Center;
118
Lehmann et al., 2011; Bucher et al., 2012).
119
120
Filename: tect20238-sup-0004-pS02.tif
121
File descriptor: Aerial view of LNS
122
Figure S2: Aerial view of the area depicted in Figure 3a. Aerial photographs were acquired with
123
the Modular Airborne Camera System, developed by the DLR (German Aerospace Center;
124
Lehmann et al., 2011; Bucher et al., 2012).
125
126
Filename: tect20238-sup-0005-pS03.tif
127
File descriptor: Profiles of selected channels
128
Figure S3: Longitudinal and cross-sectional profiles of selected channels. a) Longitudinal profiles
129
(A, B, C) of three selected main channels in the La Negra segment. The vertical line marks the
130
intersection with the surface rupture. The profiles do not contain any knickpoints related to the
131
analyzed surface rupture. b) Cross-sectional profiles of the channels in panel a). Cross-sections
132
have been extracted at the surface rupture. Channel widths have been measured across the most
6
133
deeply incised sections of the channels. c) Longitudinal profiles (D, E, F) of three gullies from
134
the Cumbre Jequier segment. Knickpoints related to the analyzed surface rupture are marked with
135
circles. d) Cross-sectional profiles of the gullies in panel c). Cross-sections have been extracted at
136
the surface rupture. Channel widths have been measured across the most deeply incised sections
137
of the channels.
138
139
Filename: tect20238-sup-0006-pS04.tif
140
File descriptor: Synthetic profiles
141
Figure S4: Synthetic profiles generated with SDL [Hilley and Arrowsmith, 2003]. Initial scarp
142
slope angle θ1 = 80°, data point spacing d = 0.1 m, background slope angle α = 0°. Consecutive
143
time steps from 0 a to 100 ka are marked with different colors. Diffusion constant for degradation
144
forward modeling κ1 = 0.46 m²/ka. a) Synthetic profile data. b) Calculated gradient data, window
145
size = 3. Inset shows zoom between -10 and +10 m. c) Calculated gradient data, window
146
size = 25. Inset shows zoom between -10 and +10 m.
147
148
Filename: tect20238-sup-0007-pS05.tif
149
File descriptor: Synthetic profiles
150
Figure S5: Synthetic profiles generated with SDL [Hilley and Arrowsmith, 2003]. Initial scarp
151
slope angle θ2 = 60°, data point spacing d = 0.1 m, background slope angle α = 0°. Consecutive
152
time steps from 0 a to 100 ka are marked with different colors. Diffusion constant for degradation
153
forward modeling κ1 = 0.46 m²/ka. a) Synthetic profile data. Inset shows zoom between -20 and
7
154
+10 m. b) Calculated gradient data, window size = 3. Inset shows zoom between -10 and +10 m.
155
c) Calculated gradient data, window size = 25. Inset shows zoom between -10 and +10 m.
156
157
Filename: tect20238-sup-0008-pS06.tif
158
File descriptor: Synthetic profiles
159
Figure S6: Synthetic profiles generated with SDL [Hilley and Arrowsmith, 2003]. Initial scarp
160
slope angle θ1 = 80°, data point spacing d = 0.1 m, background slope angle α = 0°. Consecutive
161
time steps from 0 a to 100 ka are marked with different colors. Diffusion constant for degradation
162
forward modeling κ2 = 0.046 m²/ka. a) Synthetic profile data. b) Calculated gradient data,
163
window size = 3. Inset shows zoom between -10 and +10 m. c) Calculated gradient data, window
164
size = 25. Inset shows zoom between -10 and +10 m.
165
166
Filename: tect20238-sup-0009-pS07.tif
167
File descriptor: Synthetic profiles
168
Figure S7: Synthetic profiles generated with SDL [Hilley and Arrowsmith, 2003]. Initial scarp
169
slope angle θ2 = 60°, data point spacing d = 0.1 m, background slope angle α = 0°. Consecutive
170
time steps from 0 a to 100 ka are marked with different colors. Diffusion constant for degradation
171
forward modeling κ2 = 0.046 m²/ka. a) Synthetic profile data. Inset shows zoom between -20 and
172
10 m. b) Calculated gradient data, window size = 3. Inset shows zoom between -10 and +10 m.
173
c) Calculated gradient data, window size = 25. Inset shows zoom between -10 and +10 m.
174
8
175
Filename: tect20238-sup-0010-pS08.tif
176
File descriptor: Synthetic profiles
177
Figure S8: Synthetic profiles generated with SDL [Hilley and Arrowsmith, 2003]. Initial scarp
178
slope angle θ2 = 60°, data point spacing d = 0.1 m, background slope angle α = 4°. Consecutive
179
time steps from 0 a to 100 ka are marked with different colors. Diffusion constant for degradation
180
forward modeling κ1 = 0.46 m²/ka. a) Synthetic profile data. Inset shows zoom between -20 and
181
5 m. b) Calculated gradient data, window size = 3. Inset shows zoom between -20 and +20 m.
182
c) Calculated gradient data, window size = 25. Inset shows zoom between -20 and +20 m.
183
184
Filename: tect20238-sup-0011-pS09.tif
185
File descriptor: Synthetic profiles
186
Figure S9: Synthetic profiles generated with SDL [Hilley and Arrowsmith, 2003]. Initial scarp
187
slope angle θ2 = 60°, data point spacing d = 0.1 m, background slope angle α = 4°. Consecutive
188
time steps from 0 a to 100 ka are marked with different colors. Diffusion constant for degradation
189
forward modeling κ2 = 0.046 m²/ka. a) Synthetic profile data. Inset shows zoom between -20 and
190
5 m. b) Calculated gradient data, window size = 3. Inset shows zoom between -15 and +15 m.
191
c) Calculated gradient data, window size = 25. Inset shows zoom between -15 and +15 m.
192
193
Filename: tect20238-sup-0012-pS10.tif
194
File descriptor: Synthetic SRL from moment magnitude (Mw)
9
195
Figure S10: Synthetic SRL calculated from moment magnitude (Mw) using the regression
196
function (Table 1, equation (6)) by Wells and Coppersmith [1994]. Magnitudes used for
197
calculation of SRL have been previously determined based on measured displacement (LDI, and
198
KnpD). The hatched boxes mark the range of magnitudes and the resulting synthetic SRL.
10
Download