The development of a remote sensing based technique to predict debris flowtriggering conditions in the French Alps

The effects of mass movements, including debris flows, on the inhabitants o f mountainous regions can often be catastrophic, causing serious casualties and property damage. These impacts could potentially be reduced with the d evelopment of early warning systems. Debris flows are generally initiated b y either heavy rainfall or snow melt. In the past, prediction of debris flo w events has been limited to the moment of the how onset and dependant on t he accurate description of free flowing water conditions at the time of ini tiation. This has remained problematic not least because of the high spatia l and temporal variabilities of the triggering phenomena, making their accu rate measurement by conventional means, such as by raingauges, difficult. Remote sensing data offers an ideal opportunity to provide information on d ebris how triggering conditions, including details of the evolution of trig gering rainfall conditions before they initiate a debris how event. In this paper we outline the development of a remote sensing technique to provide early warning of debris flow triggering conditions using infrared data meas ured from the Meteosat satellite series, for the Bachelard Valley in the Fr ench Alps. The relatively simple relationship and short time interval betwe en the onset of heavy rainfall, and the initiation, movement and deposition of a debris flow allows information on the triggering conditions to be con sidered as early warning of the actual debris how event itself, in location s of known debris flow hazard. Predictive information of triggering conditi ons of a particular hazard is of vital importance to the development of an effective early warning system. The technique outlined in this paper was de veloped using the debris flow initiation model, of Blijenberg et al., linke d to automated raingauges over a four year period from 1991 to 1994. Of the six case studies identified warning times of 1-12 hours were given in five of these. A false alarm test over a month period for the region revealed f alse alarms on two days, only. This paper shows that high temporal resoluti on remote sensing data can be used to provide early warning of atmospheric conditions likely to initiate debris flow events. This information is of im portance to the development of a debris flow hazard early warning system.