Keywords: climate analysis, climate impact, climate adaptation, hazard trigger patterns, railway network, hazard events.

The geo-coded hazardous event data base, whose establishment is lined out in part I of this three-paper-series – whereby we have focussed in the project 'clim_ect' on the following hazard categories: (i) flooding, (ii) mudslide, (iii), wind-storm, (iv) falling rock, (v) snow – has been blended with meteorological data (used as predictors in the modelling of hazard occurrences). This has allowed to calibrate and validate a model based on empirical orthogonal functions (EOFs) in T-mode to derive pertaining hazard trigger patterns (HTPs). A specifically designed validation procedure that is being based on a substantial amount of bootstrapped train/test subsets allowed the identification of optimal hazard-category-sensitive-parameter-settings most suitable for subsequent calculations of hazard-specific HTPs. This procedure has lead to the following model parameters: 3 EOF patterns, a temporal window of 7 days, and category-dependant predictors out of daily precipitation totals (RR), minimum daily temperature (Tn), and daily average air pressure (P) as follows (enumeration corresponds to the hazard categories from above): (i) RR, (ii) RR, (iii) P, (iv) Tn, RR, (v) Tn, RR. Derived parameter values turn out to be consistent with expert knowledge. Due to the limited scope of this short paper, two hazard categories are outlined in detail: mudslide and flooding. In case of mudslide, comparably high precipitation totals the day prior an event seem to carry decisive importance (first EOF). Target-day (the day on which a hazard event occurred) precipitation sums (second EOF) and the amount of soil-prehumidification during the week before events (third EOF) carry some significance too. In the case of flooding, the first EOF highlights the importance of continuous daily precipitation events during the week preceding events, without putting much weight on the temporal order of daily totals. The second EOF indicates the importance of pre-moistening effects. The third EOF points toward the relevance of precipitation on the day before the events emergence. Detected HTPs can be used in order to identify similar sequences in ensembles of future downscaled climate change projections, which is an extension of the described work in this paper towards an assessment of future changes in risk landscapes. Such approaches permit for quantitative assessments of developments concerning weather-driven damages that, in turn, may serve as objective foundations for setting up suitable adaption programs and protection measures.

download the full-text of this paper (PDF, 7 pages, 349 Kb)

go to the previous paper
go to the next paper
return to the table of contents
return to the volume description