Regional evaluation of three day snow depth for avalanche hazard mapping in Switzerland

摘要: Abstract. The distribution of the maximum annual three day snow fall depth H72, used for avalanche hazard mapping according to Swiss procedure (Sp), is investigated a network 124 stations in Alpine part Switzerland, using data set dating back 1931. Stationarity time investigated, showing practice no significant trend considered period. Building on previous studies about climatology Switzerland and an iterative approach based statistical tests regional homogeneity scaling H72 with altitude, seven homogenous regions are identified. A index value then developed estimate T-years return period quantiles at each single site i, H72i(T). sample average μH72i. dimensionless values H*72i=H72i / μH72i grouped one region their frequency occurrence accommodated by General Extreme Value, GEV, probability distribution, including Gumbel. proposed distributions, valid homogeneous regions, can be assess H*72i. It shown that H72i(T) estimated more accurate than calculated fitting, particularly high periods. sampling strategy accuracy also suggested value, i.e. μH72i, critical evaluation H72i. valuable because it gives input dynamics models present analysis, so decreasing uncertainty procedure.