Document Actions

IJEGE-11_BS-Pavlova-et-alii

background image
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
127
DOI: 10.4408/IJEGE.2011-03.B-015
DEBRIS FLOW OCCURRENCE AND METEOROLOGICAL FACTORS
IN THE FRENCH ALPS: A REGIONAL INVESTIGATION
i
Rina
PAVLOVA
(*)
, v
inCent
JOMELLI
(*)
, d
elPHine
GRANCHER
(*)
,
d
aniel
BRUNSTEIN
(*)
& m
atHieu
VRAC
(**)
(*)
CNRS, Laboratoire de Géographie Physique de Meudon, France
(**)
CNRS, Laboratoire des Sciences du Climat et de l'Environnement, France
number of rainy days between May and September)
correctly explains more than 60% of the DF events.
K
ey
words
: DF probability, logit, meteorological parame-
ters, regional scale.
INTRODUCTION
Debris flow (DF) is a dominant mass movement
process in mountain areas all over the world and is a
significant natural hazard. A classical distinction is gen-
erally made between a debris flood corresponding to a
rapid, surging flow of water, heavily charged with debris
in a steep channel, and a debris avalanche corresponding
to a rapid or extremely rapid shallow flow of partially or
fully saturated debris on a steep slope without confine-
ment in an established channel (H
unGR
et alii, 2008).
In a warming climate in the Alps, for instance,
regional climate models generally agree that precipi-
tation would be likely to undergo seasonal shifts and
that higher interannual variability could occur and
thus cause an increase in extreme precipitation events
(C
HRistensen
& C
HRistensen
, 2004 d
éQué
, 2007). As
it is known that debris flows and debris avalanches
are often triggered by intense summer rainy events
(C
aine
1980; i
veRson
1997; G
uzzetti
et alii, 2007), a
change in the global climate in the future could have
an impact on the magnitude and /or frequency of these
processes (J
omelli
et alii, 2007, 2009). Consequently,
a better knowledge of the relationship between debris
flow/avalanche and climate is a fundamental issue.
ABSTRACT
Debris flow (DF) phenomena is at the top of the
list of dangerous natural hazards in the mountains are-
as all over the world. Among factors resulting in a DF
triggering, meteorological conditions are considered
to be the most relevant. The general objective of this
study was to identify meteorological parameters con-
trolling the triggering of DF in one part of the French
Alps over the last 50 years.
Major factors are quite well explored at the global
scale or contrariwise in very precise territory in partic-
ular catchment areas. However for now we have a poor
knowledge of those factors at the scale of a medium-
sized region (including catchments with different geo-
morphic characteristics over several km
2
) especially in
the French Alps. In addition in this region only a few
studies focused on relationships with climate.
To understand DF activity and link it with mete-
orological parameters in the north region of the French
Alps, we used a multivariate statistical approach. Re-
gional meteorological parameters (such as mean month-
ly temperature and precipitation) were first computed
from a Principal Component Analysis of observed me-
teorological data from four weather stations. A binomial
monthly logistic regression (LR) probability model was
then fitted between the main principal components and
DF data base composed of 298 debris flow events trig-
gered between 1971 and 2008. Results revealed that
the most successful model including two meteorologi-
cal predictors (minimal monthly temperature and the
background image
I. PAVLOVA, V. JOMELLI, D. GRANCHER, D. BRUNSTEIN & M. VRAC
128
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
In between the two valleys there is a huge mountain
massif de la Vanoise with Grande Casse peak (3855 m
asl) and a mean altitude of around 2 000 m asl.
Geologically, the SE part of this region including
the upper part of the valley d’Arc mainly consists of
schistes, while the central massive de la Vanoise is a
crystalline outbreak and the lowest NW part is cov-
ered by sedimentary rocks.
This region is one of the most popular tourist areas
of the French Alps with many famous ski resort stations
that receive more than one million tourists each season.
Buildings and road networks are potentially threatened
by debris flows hazards. There is therefore a growing
demand for hazard zoning and revision of DF protection.
DEBRIS FLOw EVENTS DATABASE
Data used in this study is from a survey of DF
triggering conducted by the service Restauration des
Terrains de Montagne (RTM) which was initiated in
the 1900s by foresters and covers the entire French
Alps. The data is organized by a department and for
our study we used the database of Savoie department.
DF data for the Savoie region were collected from
scientific and technical journals, monographs by local
publishers, technical reports and unpublished docu-
ments gathered from the archives of local authorities
and state agencies stored at RTM services.
Over the last few decades several papers have
been published on the links between climate condi-
tions and triggering of debris flows. These links were
investigated at two distinct spatial scales. Some studies
analyzed the climatic causes of debris flow triggered
in a few small catchment areas with an accurate docu-
mentation of geomorphic characteristics and meteoro-
logical stations located in the catchment area (G
oRse
-
vski
et alii 2000; s
toffel
and
b
eniston
, 2006; R
uPeRt
et alii 2008). Other studies investigated the impacts of
climate conditions on debris flows at large spatial scale
by focusing on a regional trend over a region of sev-
eral thousand km
2
(H
aebeRli
et alii, 1990; z
immeRman
,
1990; R
ebetez
et alii, 1997; G
uzzetti
et alii, 2007).
Such spatial scale investigations were mainly based on
statistical modeling rather than determinist analyses.
In the French Alps, very few studies have been
conducted at a large spatial scale (several km
2
). Most
studies were composed of technical reports that fo-
cused on a special catchment area that revealed risks
of DF. J
omelli
et alii, (2003, 2004, 2007) analyzed the
relationships between climate conditions and different
types of debris avalanches as a function of their lithol-
ogy or of the nature of the accumulated debris over a
part of the southern French Alps. In most cases, ex-
treme precipitation in summer and the number of freez-
ing days were considered as significant parameters.
At a smaller spatial scale, (R
emaitRe
, 2006) analyzed
climate conditions responsible for the triggering of 12 de-
bris flows in one valley of the southern French Alps. He
did not observe any significant relationship between an-
nual precipitation and the triggering of debris flows. By
contrast he identified two causes that may be responsible,
namely i) daily total precipitation above 20 mm and ii)
monthly mean precipitation above the mean value.
The main goal of this paper is to document the re-
lationships between current climate and DF activity in
the northern part of the French Alps, based on a large
number of debris flow events selected in a large territory.
GENERAL SETTING
STUDY AREA
The territory considered in this paper corresponds
to the department of the Savoie located in the North
Alps (Fig.1). It contains two large river valleys (both
of which are NW-SE oriented) the upper Isere and
Grand Arc valleys and their effluents. These huge prin-
cipal valley catchment areas cover more than 300 km
2
.
Fig. 1 - DF catchment area and location of the meteoro-
logical stations
background image
DEBRIS FLOW OCCURRENCE AND METEOROLOGICAL FACTORS IN THE FRENCH ALPS: A REGIONAL INVESTIGATION
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
129
For further statistical treatments, additional fac-
tors were calculated for each station such as the
monthly mean precipitations for instance from May to
September. We added monthly number of rainy days
and the number of days between May and Septem-
ber during which rainy events were greater than 10,
20 and 30 mm/day. Temperature factors consisted of
monthly mean minimal and maximal temperature.
METHODS
REGIONAL CLIMATE CONDITIONS
We used data collected from the four meteorologi-
cal stations to compute regional climate conditions
that were used to document the relationship between
climate and debris flows activity. For our purpose we
used Principal Components Analysis to extract the
main part of the variance of the regional climate signal
from the four different meteorological stations.
This statistical analysis also allowed us to reduce
data dimensionality (J
olliffe
, 2002). By using Princi-
pal Components analysis separately for each tempera-
ture and precipitation characteristics principal com-
ponents (PCs) were generated then used for further
analysis in logistic regression probability model.
LOGISTIC REGRESSION MODEL
Logistic regression (LR) analysis is often used
to investigate the relationship between a set of ex-
planatory variables such as meteorological factors and
discrete responses like event/non-event or presence/
absence (H
osmeR
& l
emesHow
, 2000).
The logistic regression estimates probabilities of
the event and non-event occurrence, conditionally on
the explanatory variables. LR is based on the logit
function defined as
where p is probability to be estimated. The main
assumption is that logit (p) can be approximated by a
linear combination of the explanatory variables:
Descriptors of each DF event include an identifica-
tion number for each DF catchment area, date and time of
the event, elevation of the starting and runout zone, some
characteristics defining DF morphometrics, meteorologi-
cal observations during and up to three days before the
event, human casualties, and damage to infrastructure.
Although, the quality and reliability of the sur-
veys vary from site to site, the record concerning
Savoie department is considered as one of the best by
the service RTM. Because DF survey methods have
changed significantly over the years, our statistical
analysis was conducted over a 37-year period (1971-
2008) which features high-quality records and greater
homogeneity of survey methods.
One problem was to distinguish DF from other
natural hazards like floods. For that reason, we only
reviewed the active period for DF from May to Sep-
tember and carefully checked the description of the
events made by the forest ranger.
The analysis of old documents was completed by
a field trip where the most active DF catchment areas
were visited. Since the spring 1971 298 daily dated
DF events from 102 catchments have been recorded
in the Savoie region.
Mean DF catchment area in the region is around 3
km
2
with talweg length of around 1 km. The DF altitude
zone is between 800-2800 m. All the catchment areas
selected in this study are quite small with less than 2
km
2
mean surface areas, but very well surveyed because
they are located close to the railway road network.
Twenty-eight debris flows that were impossible to
date were not taken into account in the analysis. Half of
the recorded DF catchment areas had DF triggering event
only once in the current period. In the other half, DF
event ranged from two to five and only five catchment ar-
eas were very active with more than 20 recorded events.
METEOROLOGICAL DATA
To characterize climatic conditions in the studied
region, we selected four meteorological stations at
different locations and elevations (Table 1). Observed
cumulative precipitation and minimal and maximum
temperature data at a daily time scale were available
for four meteorological stations in the North Alps re-
gion for the period 1971-2008. These Météo-France
stations are situated in Beaufort, Bourg st Maurice,
Pralognan and Lanslebourg villages and are equally
distributed over the investigated territory (Fig. 1).
Tab. 1 - Meteorological stations with daily precipitation
and temperature data for 1971-2008 period
(1)
background image
I. PAVLOVA, V. JOMELLI, D. GRANCHER, D. BRUNSTEIN & M. VRAC
130
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
After all different tests, the best compilation of
temperature and precipitation parameters was chosen
based on the highest LR model coefficients values.
RESULTS
During the period of DF activity (from May to
September and from 1971 to 2008), there were 98
months without triggering and 88 months with DF ac-
tivity in the region.
Primarily parameters used as explanatory factors for
the triggering of DF were: mean monthly precipitation,
number of rainy days per month, number of rainy days
with daily rain cumulative greater than 10, 20 and 30
mm/day, monthly minimal and maximum temperatures.
PCs analysis was then conducted on these param-
eters to compute independent regional climate param-
eters. For precipitation parameters more than 70%
common signal is documented from the first compo-
nent (F1) and 90% of common temperature compo-
nent series. For that reason only this component was
retained for further analysis (Tab.2).
The F1variables contribution consisting of data
of four chosen meteorological stations is nearly equal
(Tab.3). The weight of each meteorological station
in PC’s F1 is around 25% and only Lanslebourg has
lower meanings for precipitation parameters.
We first tested all F1 parameters one by one to
with x
k
the k
th
predictor, a
k
the k
th
linear coefficient and
k the number of predictors.
By simply inverting Eq. (1), we have that
or equivalently
which, based on Eq. 2, can be written as:
Eq. (5) is used in practice to relate the probability
p of an event with the meteorological factors (x
k
)
k=1,…,k
Our objective for the LR probability model at a
monthly time scale was to find the best temperature
and precipitation parameters that would explain DF
triggering in the region.
The LR model was run monthly (from May to
September), by assigning the label 1 to months in-
cluding a DF triggering and the label 0 to months in
which DF did not occur. The series of first PCs for
each meteorological parameter were tested as inde-
pendent explicative variables.
To check the quality of the model several verifica-
tion tests for each LR were computed and then com-
pared to select the most significant model results.
First the probability of the adjusted model was
tested against a test model. If this probability Pr >
LR is less than the 0.05 significance threshold that
was set, then the contribution of the variable to the
adjustment of the model is significant. Otherwise, it
can be removed from the model.
Next the estimated values is Wald's Chi
2
test,
where the likelihood-ratio test is performed in which
(θ-θ
0
)
2
/var(θ) is compared to a chi-square distribution,
where the maximum likelihood estimate θ of the pa-
rameter of interest θ, is compared with a suggested
value θ
0
, which is zero in our case. The best model
should have the greatest Chi
2
values.
The table of standardized coefficients was used
to compare the relative weights of the variables. The
higher the absolute value of a coefficient, the more
important the weight of the corresponding variable.
When the confidence interval around standardized
coefficients is null, the weight of a variable in the
model is not significant.
(5)
(4)
(3)
(2)
Tab. 2 - PCA functions for meteorological parameters
Tab. 3 - weight of four meteorological stations in PC’s
functions interior distribution of meteorological
parameters
background image
DEBRIS FLOW OCCURRENCE AND METEOROLOGICAL FACTORS IN THE FRENCH ALPS: A REGIONAL INVESTIGATION
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
131
distinguish the most significant parameters in the tem-
perature and precipitation series.
After trying different combinations two distinct
significant models were determined. The LR models
that either only used the number of rainy days (Nrd)
between May and September 1971-2008 or monthly
minimal temperature (Tn) explained 60 and 62% of
triggering respectively (Tab. 4).
Numbers of rainy days with daily rain cumulative
greater than 10, 20 and 30 mm/day were expected to
be closely correlated with DF events as indicators of
extreme precipitation events, but the tests revealed
that these extreme precipitation parameters were less
or insignificant in our case. Supposing that PCA ap-
proach may be less relevant to extreme precipitation
events, tests were conducted with observed data but
they did not revealed better results.
As the monthly minimal temperature (Tn) and the
number of rainy days (Nrd) between May and September
1971-2008 were the best parameters analyzed separately,
we computed another LR model which combined the two
best parameters. The model has a greater significance and
higher coefficient values than two parameters independ-
ently taken (Table. 5a). The percentage of correct predic-
tions for presence/absence of DF event was greater than
60% (Table. 5b). Tests, conducted with other combina-
tions of climate parameters gave less significant results.
The LR probability model linear equation coming
from equations (1) and (2) is:
Logit (p1) = -0.14 + (0.44*Tn) + (0.39*Nrd)
Tab. 4 - Statistical parameters of two independent logis-
tic models for most significant meteorological
parameters
Figure 2 shows the probability of a DF being trig-
gered as a function of the monthly minimal temperature
(Tn) and number of rainy days (Nrd) per month between
April and October. The probability of a DF increased
considerably when bought F1 of number of rainy days
and F1 for temperature parameter was positive. The
probability of a DF event greater than 0.8 was observed
when the F1 of Nrd was greater than 0 and the F1 of Tn
was greater than 2. The lowest probability values (0.2-
0.4) were observed with negative F1 of Nrd and Tn.
DISCUSSION AND CONCLUDING REMARKS
DOES THE QUALITY OF THE DF SURVEY IN-
FLUENCE THE MODELS?
The national natural hazard survey documented
more than 900 debris flows in the whole French Alps
and 298 were surveyed in the Savoie department over
the last four decades. For several reasons, the qual-
ity of surveys changed during these last decades and
conditions for surveys in the Savoie department were
improved after 1971 because of the rise of the compe-
tence and motivation of the observers. Nevertheless,
the number of debris flows used in this study was kept
to the minimum. Second, discrepancies may exist be-
tween the date of any debris flow event and meteoro-
logical data computed in the model. This means that the
data used for a given debris flow event may be younger
than the meteorological data used in the dataset by one
or two days. However, by comparing probabilities cal-
culated for dates of the triggering of debris flows with
those calculated for days preceding any debris flow
event, we estimated that in most cases (86%), prob-
abilities were lower for the day preceding a given event.
Tab. 5a (top). LR model output using the best temperature
and precipitation parameters: Nrd (number of
rainy days) and Tn (minimal temperature). 5b
(bottom). Percentage of correct predictions of DF
events: presence - 1, absence 0
(6)
Fig. 2 - DF triggering probability: Nrd= F1 for number of
monthly rainy days. Tn= F1 for monthly minimal
temperature
background image
I. PAVLOVA, V. JOMELLI, D. GRANCHER, D. BRUNSTEIN & M. VRAC
132
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
the day as the DF was triggered, which may explain why
extreme precipitation events only were less or not signif-
icant in our analysis. Moreover tests conducted during
the hottest summer period did not reveal better results.
It is well known that two conditions required for de-
bris flow to be triggered (C
aine
, 1980; J
oHnson
& R
odine
,
1984; v
an
s
teiJn
, 1996): heavy rainfall of long duration
or high intensity as mentioned above and a large volume
of debris. The relationship between the triggering of a de-
bris flow and the minimum monthly temperature may also
suggest such a relationships with the stock of rock debris
as observed in other parts of the French Alps (J
omelli
et
alii, 2004). The volume of debris is due either to morainic
accumulations on a slope (H
aebeRli
et alii, 1990) or to
debris accumulated in a catchment area or at the apex of a
slope deposit (P
eCH
& J
omelli
, 2001) by frost weathering
or by erosion in the track (J
akob
et alii, 2005).
The increase in temperatures at high altitudes over
the last few decades has clearly demonstrated by differ-
ent done on homogenized series (b
eniston
et alii, 1997;
d
iaz
& b
Radley
, 1997; b
öHm
et alii, 2001) as well as at
the four stations used in this paper. The mean altitude of
the triggering zone of debris flows is about 2200 m in this
region close to the 0°C isotherm. The temperature un-
derlined in the model could at least in some cases reflect
a possible degradation in the permafrost either in rock
walls as has been observed elsewhere in the Alps (H
ae
-
beRli
et alii, 1993; H
aebeRli
& b
eniston
, 1998; w
eG
-
mann
et alii, 1998; i
mHof
et alii, 2000) or in the track.
Another possible interpretation is linked to the influence
of temperature on snow/rain limit with liquid precipi-
tation as rain occurring at high altitudes. It could also
reflect possible changes in the duration of snow cover
and expose high slopes to greater temperature variations.
Further detailed studies will be conducted to explore
these different possible explanations proposed here and
analyze possible differences between the northern and
the southern parts of the French Alps.
ACKNOWLEDGEMENTS
This research was investigated by the Laboratory of
Physical Geography (LGP, CNRS-Meudon) in the frame-
work of two projects ACQWA (Assessing Climate im-
pacts on the Quantity and quality of Water) and Scampei
(Scenarios Climatiques Adaptes aux zones des Montagne:
Phenomenes extremes, Enneigement et Incertitudes). Spe-
cial thanks to the Restauration des Terrains de Montagne
in Savoie region for allowing us to work on their data base.
To test the sensitivity of our models, bootstrap analyses
were performed. They showed that the mean of the errors
between parameters obtained from each simulated sample
and those obtained from the original dataset was low, dem-
onstrating that these parameters are statistically stable.
THE SIGNIFICANCE OF THE METEOROLOGI-
CAL PARAMETERS USED IN THE MODELS
The statistical analysis of debris-flow inventory in
the Savoie region allows us to discuss the role of the
different climatic parameters in the triggering of debris-
flows. The models presented here are in agreement
with current data on debris flow dynamics. In Iceland
(d
eCaulne
& s
aemundson
2007), the U.S. (C
annon
et
alii, 2008), and the Swiss Alps (H
aebeRli
et alii, 1990;
z
immeRmann
& H
aebeRli
, 1992; R
ebetez
et alii, 1997;
z
immeRmann
, 2005) it was shown that the probability
of debris flow occurrence correlated primarily with pre-
cipitation and most often with total precipitation for the
15 days preceding a given debris flow event or intense
rainy events. Our model revealed that debris flow trig-
gering in this part of the French Alps underlines pre-
cipitation as a significant parameter as well.
It is also interesting to note that our model points
to temperature as a significant parameter in the trig-
gering of debris flows. There are several possible in-
terpretations of the significance of this parameter.
Tests revealed that extreme precipitation between
10 and 20 mm/day may be a significant parameter if it is
combined with temperature. The probability of DF trig-
gering driven by the number of rainy days greater than
10, 20 mm combined with monthly minimal and maxi-
mum temperature explained lower than 50% of trigger-
ingless than the best model presented in the previous
paragraph. Tests using the number of days during which
rainy events were greater than 20 mm/day resulted in
small insignificant correlations. The possible explana-
tion for this relationship with extreme precipitation is
quite complex. On one hand, extreme precipitations
with high values are quite rare phenomena and they do
not occur every month, so there are lots of zero values
as input data for LR analysis. On the other hand, in sum-
mer such storms occurring during hot days are often
very localized and as such are not always recorded by
the rain gauges at weather station. Simple comparisons
of data concerning DF events and rains recorded by the
nearest weather station showed that in less than 50 %
cases cumulated rain lower than 10 mm/d was observed
background image
DEBRIS FLOW OCCURRENCE AND METEOROLOGICAL FACTORS IN THE FRENCH ALPS: A REGIONAL INVESTIGATION
Italian Journal of Engineering Geology and Environment - Book www.ijege.uniroma1.it © 2011 Casa Editrice Università La Sapienza
133
REFERENCES
b
eniston
m., d
iaz
H. f. & b
Radley
R. s. (1997) - Climatic change at high elevation sites: an overview. Climatic Change, 36: 233-25.
b
oHm
R., a
ueR
i., b
Runetti
m., m
auGeRi
m., n
anni
t. & s
CHoneR
w. (2001) - Regional temperature variability in the European
Alps: 1760-1998 from homogenized instrumental time series. Int. J. Climatol., 21: 1779-1801.
C
aine
N. (1980) - The rainfall intensity-duration control of shallow landslides and debris-flows. Geografiska Annaler, 62A (1-2): 23-27.
C
annon
s.H., G
RtHeR
J. e., w
ilson
R. C., b
oweRs
J.C. & l
abeR
J. l. (2008) - Storm rainfall conditions for floods and debris
flows from recently burned areas in southwestern Colorado and southern California. Geomorphology, 96 (3-4): 250-269.
C
HRistensen
o.b. & C
HRistensen
J.H. (2004) - Intensification of extreme European summer precipitation in a warmer climate.
Glob Planet Change, 44:107-117.
d
eQue
(2007) - Frequency of precipitation and temperature extremes over France in an anthropogenic scenario: model results
and statistical correction according to observed values. Glob Planet Change 57: 16-26.
d
eCaulne
a. & s
aemundsson
(2007) - Spatial and temporal diversity for debris-flow meteorological control in subarctic oceanic
periglacial environments in Iceland. Earth Surface Processes and Landforms, 32: 1971-1983.
d
iaz
H. f. & b
Radley
, R. s. (1997) - Temperature Variations during the Last Century at High Elevation Sites. Clim. Change,
36 (3-4): 253-279.
G
oRsevski
P. v., R
andy
P. G. & f
oltz
b. (2000) - Spatial Prediction of Landslide Hazard Using Discriminant Analysis and GIS.
GIS in the Rockies 2000 Conference and Workshop .Applications for the 21st Century. Denver, Colorado.
G
uzzetti
f., P
eRuCCQCCi
s., R
ossi
m. & s
taRk
C.P. (2007) - Rainfall thresholds for the initiation of landslides in central and
southern Europe. Meteorol Atmos Phys, 98: 239-267.
H
aebeRli
w., R
iCkenmann
d., z
immeRmann
m. & R
ossli
u. (1990) - Investigation of 1987 Debris Flows in the Swiss Alps:
General Concept and Geophysical Soundings. IAHS Publication, 194: 303-310.
H
aebeRli
w., G
uodonG
C., G
oRbunov
a. P., & H
aRRis
s. a. (1993) - Mountain Permafrost and Climatic Change. Permafrost
Periglacial Proc. 4 (2): 165-174.
H
aebeRli
w. & b
enistin
m. (1998) - Climate change and its impacts on glaciers and permafrost in the Alps. Ambio 27: 258-265.
H
osmeR
d. & l
emesHow
s. (2000) - Applied Logistic Regression, 2nd Edition Wiley.
H
unGR
o., m
CdouGall
s., w
ise
m. & C
ullen
m. (2008) - Magnitude-frequency relationships of debris flows and debris
avalanches in relation to slope relief. Geomorphology, 96 (3-4): 355-365.
i
mHof
m., P
ieRRenHumbeRt
G., H
aebeRli
w. & k
ienHolz
H. (2000) - Permafrost investigation in the Schilthorn massif, Bernese
Alps, Switzerland. Permafrost and Periglacial Processes 11: 189-206.
i
veRson
R. M. (1997) - The physics of debris flows. Rev. Geophys., 35 (3): 245-296.
J
abob
, b
ovis
& o
den
(2005) - The significance of channel recharge rates for estimating DF magnitude and frequency. Earth
surface processes and landforms, 21: 755-766.
J
olliffe
I.T. (2002) - Principal Component Analysis. 522, Second Edition. Springer, New York.
J
omelli
v., C
HoCHillon
C., b
Runstein
d. & P
eCH
P. (2003) - Hillslope debris flows occurrence since the beginning of the 20
th
century in the French Alps. In: Debris flow hazards mitigation. 127-137, Millpress, Rotterdam.
J
omelli
v., P
eCH
P., C
HoCHillon
C. & b
Runstein
d. (2004) - Geomorphic variations of debris flows and recent climatic change
in the French Alps. Climatic Change, 64: 77-102.
J
omelli
v., b
Runstein
d., G
RanCHeR
d. & P
eCH
P. (2007) - Is the response of hill slope debris flows to recent climate change
univocal? A case study in the Massif des Ecrins (French Alps). Climate Change, 85:119-137.
J
omelli
v., b
Runstein
d., v
RaC
m., d
eQue
m. & G
RanCHeR
d. (2009) - Impacts of future climatic change (2070-2099) on the
potential occurrence of debris flows: a case study in the Massif des Ecrins (French Alps). Climate Change DOI 10.1007/
s10584-009-9616-0.
J
oHson
a.m.& R
odine
J.R. (1984) - Debris flow, in b
Rundsen
, d., & P
RioR
, D.B., eds., Slope instability: Chichester, England,
John Wiley & Sons. 257-361.
P
eCH
P. & J
omelli
v. (2001) - Caractéristiques et rôle du cône apical dans le déclenchement des coulées de débris. Géographie
physique et Quaternaire, 55 (1): 47-61.
R
emaitRe
A. (2006) - Morphologie et dynamique des laves torrentielles: Application aux torrents des terres noires du bassin de
Barcelonnette. PHD Thesis, 483.
background image
I. PAVLOVA, V. JOMELLI, D. GRANCHER, D. BRUNSTEIN & M. VRAC
134
5th International Conference on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment Padua, Italy - 14-17 June 2011
R
ebetez
m., l
uGon
R. & b
aeRiswyl
P.a. (1997) - Climatic change and debris flows in high mountain regions: the case study of
the Ritigraben torrent (Swiss Alps). Climatic Change 36 (3/4): 371-389.
R
uPeRt
m.G., C
annon
s.H., G
aRtneR
J.e. m
iCHael
J.a. & H
elsel
d.R. (2008) - Using logistic regression to predict the probability
of dfs in areas burned by wildfires, southern California. 2003-2006: U.S. geological survey open-file report 1370, 9.
s
toffel
m. & b
eniston
m. (2006) - On the incidence of debris flows from the early Little Ice Age to a future greenhouse climate:
a case study from the Swiss Alps. Geophysical Research Letters 33: L16404
v
an
s
teiJn
H. (1996) - Debris-flow magnitude-frequency relationships for mountainous regions of Central and Northwest
Europe. Geomorphology 15 (3-4): 259-273.
w
eGmann
m., G
udmudsson
G. H. & H
aebeRli
w. (1998) - Permafrost changes in rock walls and the retreat of Alpine glaciers:
A thermal modelling approach. Permafrost Periglac., 9: 23-33.
z
immeRmann
M. (1990) - Debris Flows 1987 in Switzerland: Geomorphological and Meteorological Aspects. , Proceedings of
two Lausanne Symposi, IAHS Publication, 194: 387-393.
z
immeRmann
m. & H
aebeRli
w. (1992) - Climatic Change and Debris Flow Activity in High- Mountain Areas - A Case Study in
the Swiss Alps. Greenhouse-Impact on Cold-Climate Ecosystems and Landscapes, Catena Suppl. 22: 59-72.
z
immeRmann
M. (2005) - Analysis and management of debris flow risks at Sörenberg (Switzerland). In: J
akob
m., H
unGR
o.
(2005, eds.) - Debris-Flow Hazards and Related Phenomena., 615-634, Springer, Berlin
Statistics