10.3 Experimental Methods
155
by the authors themselves, and its performance was confirmed in the calibration tests
and the landslide experiments (Xu et al. 2015a, b).
Gravitational erosion involved both large-scale mass wasting and smaller-scale
erosion. The size of each mass failure was calculated and classified, and then the
total amount of all failure masses g t and the peak value of individual erosion events
during a rainfall event g p were obtained. All failure masses with a volume of more
than 500 cm
3 were considered in the experimental study. To assess the effects of the
initial landform geometry on the gravity erosion, we divided experiments into the
following eight experimental groups, each of which had the same slope height or
gradient:
G1 (Experiments L5–8) versus G2 (Experiments L1–4). Rainfall intensity in the
former experimental group was 0.8 mm/min, while the later was 2.0 mm/min.
G3 (Experiments L9–10) versus G4 (Experiments L5–6). Rainfall duration in the
former experimental group was 30 min, while the later was 60 min.
G5 (Experiments L1, 3, 5 and 7) versus G6 (Experiments L2, 4, 6 and 8). Slope
gradient of the initial lower slope in the former experimental group was 70°, while
the later was 80°.
G7 (Experiments L1, 2, 5 and 6) versus G8 (Experiments L3, 4, 7 and 8). Slope
height of the initial lower slope in the former experimental group was 1.0 m, while
the latter was 1.5 m.
For all groups, the maximum of the individual failure masses and the average
value of the total erosion of every experiment were compared. An advantage of the
above method is that influences caused by the randomness of gravity could be handily
overcome.
Then we used the increase-rate-analysis method to evaluate variations in the gravity erosion with respect to changes in other causal factors such as rainfall intensity
and duration, and slope gradient and height. Our sensitivity analysis focuses on g t
and g p . The increase ratio of gravity erosion R g (%) is shown as follows:
R g = ( ¯
g 2 − ¯
g 1 )/ ¯
g 1
(10.1)
where ¯
g 1 is the average value of g t or the maximum value of g p before the triggering
element was changed in an experimental group, cm
3 , and ¯
g 2 is that after the triggering
element was changed, cm
3 . An increased ratio of the above output in percent will be
calculated with the growth rate of a parameter while other parameters are fixed in an
experimental group:
S = R g /R t
(10.2)
where S is the sensitivity parameter to analyze the sensibility of the failure volume
to the triggering elements; R t = (t 2 − t 1 )/t 1 is the increased ratio of the triggering
element, namely rainfall or landform, in which t 1 is the value before being changed in
an experimental group, and t 2 is that after being changed. R t is conveyed in percent.
This approach allows for a qualitative investigation of the effect of conditioning
factors.
155
by the authors themselves, and its performance was confirmed in the calibration tests
and the landslide experiments (Xu et al. 2015a, b).
Gravitational erosion involved both large-scale mass wasting and smaller-scale
erosion. The size of each mass failure was calculated and classified, and then the
total amount of all failure masses g t and the peak value of individual erosion events
during a rainfall event g p were obtained. All failure masses with a volume of more
than 500 cm
3 were considered in the experimental study. To assess the effects of the
initial landform geometry on the gravity erosion, we divided experiments into the
following eight experimental groups, each of which had the same slope height or
gradient:
G1 (Experiments L5–8) versus G2 (Experiments L1–4). Rainfall intensity in the
former experimental group was 0.8 mm/min, while the later was 2.0 mm/min.
G3 (Experiments L9–10) versus G4 (Experiments L5–6). Rainfall duration in the
former experimental group was 30 min, while the later was 60 min.
G5 (Experiments L1, 3, 5 and 7) versus G6 (Experiments L2, 4, 6 and 8). Slope
gradient of the initial lower slope in the former experimental group was 70°, while
the later was 80°.
G7 (Experiments L1, 2, 5 and 6) versus G8 (Experiments L3, 4, 7 and 8). Slope
height of the initial lower slope in the former experimental group was 1.0 m, while
the latter was 1.5 m.
For all groups, the maximum of the individual failure masses and the average
value of the total erosion of every experiment were compared. An advantage of the
above method is that influences caused by the randomness of gravity could be handily
overcome.
Then we used the increase-rate-analysis method to evaluate variations in the gravity erosion with respect to changes in other causal factors such as rainfall intensity
and duration, and slope gradient and height. Our sensitivity analysis focuses on g t
and g p . The increase ratio of gravity erosion R g (%) is shown as follows:
R g = ( ¯
g 2 − ¯
g 1 )/ ¯
g 1
(10.1)
where ¯
g 1 is the average value of g t or the maximum value of g p before the triggering
element was changed in an experimental group, cm
3 , and ¯
g 2 is that after the triggering
element was changed, cm
3 . An increased ratio of the above output in percent will be
calculated with the growth rate of a parameter while other parameters are fixed in an
experimental group:
S = R g /R t
(10.2)
where S is the sensitivity parameter to analyze the sensibility of the failure volume
to the triggering elements; R t = (t 2 − t 1 )/t 1 is the increased ratio of the triggering
element, namely rainfall or landform, in which t 1 is the value before being changed in
an experimental group, and t 2 is that after being changed. R t is conveyed in percent.
This approach allows for a qualitative investigation of the effect of conditioning
factors.