Debris flow types, geomorphological and climatic parameters, models, Massif des Ecrins

Meteorological data have been collected from 9 stations with different elevations and periods of observation located in the Massif des Ecrins to characterize climatic change in the massif as a whole.

For granite debris flows without a glacier model, the analysis was based on 79 debris flow systems. The independent variables with the best fit are the cumulated number of days of frost since the previous event and the number of daily events with more than 30mm of rainfall between 15/06 and 15/10.

For sandstone debris flows without a glacier, the analysis was based on 29 debris flow systems. The inclination of the rock layer has a strong influence on triggering probability in cases with a limited number of days of intense precipitation.

For granite debris flows with a glacier, the analysis was based on 196 debris flow systems. When large valley glaciers were not included, an effect was observed of cumulated mean summer temperature of the 5 years preceding the triggering of an event. If all types of glaciers were included, the independent variables that fit best are the presence of a geological fault in the axis of the debris flow system and daily rainfall of more than 35mm.

For sandstone debris flows with a glacier, the data set has been made up of 23 debris flow systems. The independent variable with the best fit is the number of rainfall events with more than 30mm/day between 15/06 and 15/10. The probability of triggering is around 0.3 if there are 3 rainy events with more than 30mm rainfall in one summer and around 0.6 for 5 events.

The analysis of the return period rate (time between 2 events) in a given system has shown marked variations with respect to specific local factors. Naturally average return periods of debris flows are between 17 years (morainic accumulation) and 25 years (frost action). The influence of the presence of faults is clear. In La Selle valley, for example, the return time varied between 13 and 36 years depending on whether or not the system is located on a fault. This factor can be combined with that of snow. On the northern slope of the same valley, the return period rate of debris flows for systems without glaciers but where perennial snow packs have occurred since the 1950s is 3 times higher than elsewhere. On the other hand, in the sandstone area in the south of the Massif, it is the inclination of the rock layers that determines the activity of local systems. The average return period rate can be doubled depending on whether or not the angle of the rock inclination is conform. Together these factors result in a specific response for each valley, and these responses are sometimes even opposed, whereas the climatic conditions can be regarded as homogeneous.

Debris flows were selected from an area of 100 km² covering 12 valleys in the Massif des Ecrins. All the debris flow deposits are located between 1800m asl and 2700m asl. The 0°C annual isotherm is between 2400m (north) and 2600m (south).

647 debris flows have been analysed and dated yearly by combining 4 approaches: analysis of aerial photographs, field observations since 1995, old documents and dendrochronology. The length of each dated debris flow has been measured in the field and the altitude of the starting zone estimated.

The evolution of daily precipitation occurring between 15/06 and 15/10 has been analysed by using thresholds. For a specific threshold, a distribution of the extreme events above it has been computed and the parameters of the distribution estimated for the period 1960-1980 has been compared with those obtained for the period 1980-2000. The Generalized Pareto Distribution (GPD) was then fitted with daily precipitation. A Poisson model has been used to compare the numbers of intense precipitations per year before and after 1980. The following climatic predictors were used: seasonal and monthly average precipitation in the year the debris flow was triggered; number of days between 15/06 and 15/10 with more than 20, 25, 30, 35 and 50 mm rainfall respectively; annual and seasonal average temperature in the year an event was triggered; minimal temperature during winter or spring and total number of days of frost per year since the previous event.

A model has been made for each type of debris flow because it is controlled by a specific combination of climatic and geomorphological factors. The likelihood-ratio statistic tests have shown that the 4 models are statistically significant with a likelihood ratio between 63 and 74, and between 79 and 89% of predictions are correct.

(4) - Remarques générales