Rainfall as a landslide triggering factor an overview of recent international research

Abstract

Rainfall is the most common cause of landslides. The cost of Rainfall Triggered Landslides (RTL) is not well documented and often unobtainable. In areas where they do not pose a threat to life, great damage is caused to farmland and communication infrastructures and pasture bio-mass production is heavily reduced (Table 1). In Japan more than 10,000 RTL are reported every year which claim the lives of some 400 persons (Fukuoka, 1980); A single event has killed 100 persons and inflicted property damages estimated at 300 billion yen Shimizu (1988). Given the importance of the topic, some 138 papers dealing with RTL were selected and key information was collected in a database. Nearly 82 % of all records are local investigations carried out in 23 countries. Italy provides the largest sample as for authors' nationality and widespread proneness to landslides; followed by United States (15 %), Hong Kong (8 %), Japan and United Kingdom. About 21 % of all selected papers are methodological research or syntheses and comparisons of different methodologies. Investigations refer to widespread landsliding (69 % of AD) (that is the Available Data number for each database field), the remainder corresponds to single or few landslides. A landslide classification proposed by Hutchinsons (1995) and based on the maximum depth of failure (Vm), is adopted in this work. About 40 % of AD are intermediate or deep-seated landslides (Vm > 10 m), that include all reactivations, the rest are shallow or superficial landslides, generally first-time movements. The most frequent types are: flows, translational and rotational slides, slips, avalanches and creep, decreasing order (Cruden & Varnes 1996) with soil or debris generally constituting the landslide bodies. Daily (54 % of AD), hourly (28 %), monthly (15 %) and yearly (4 %) rainfalls are used as input. Roughly 49 % consider cumulative rainfall (a rain water height obtained adding regularly monitored rainfall) of different duration. The prevalent approach is empirical, statistical or hydrological-qualitative; one out of four is partially physical and often uses numerical modelling. About 10 % combine rainfall effect characterisation with geotechnical stability analysis. The results of research in progress have been summarised.